def get_c_ore(self, in_can_info): # print "input", in_can_info # print in_can_info[0].pos t_c_order = [] for ci in range(len(in_can_info)): tm_tar_pos = in_can_info[ci].pos tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) obs_r = [] for i in tm_obs_pos: obs_r.append(0.035) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) while 1: tm_tar_pos = in_can_info[ci].pos tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) obs_r = [] for i in tm_obs_pos: obs_r.append(0.035) for i in ore_order: if i != 'T': tm_obs_pos[i] = [4.0, 0.0] tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order if tmp_order[0] != 'T': # print "not ok" t_c_order.append([]) break # if tmp_order[0] == -1: # print "no path" # t_c_order.append([]) # break # if len(ore_order) > len(self.obs_pos): # print "no path" # t_c_order.append([]) # break # print "tricky environment for c_ore so extend", ore_order # ore_order.pop() # print "delete target", ore_order # ore_order.extend(tmp_order) # print "to", ore_order else: if tmp_order[0] == 'T': # print "ok", ore_order ore_order.pop() if len(tm_obs_pos) in ore_order: # print "\n\nThere is target!! warning!!!\n\n" t_c_order.append([]) else: t_c_order.append(ore_order) break return t_c_order
def update_env(self, in_obs_pos, in_obs_grid): self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(in_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) self.ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # self.ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order # if tmp_order[0] == 'T': # print "environment setting OK" self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(in_obs_grid) self.obs_re_pos = copy.deepcopy(in_obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop()
def set_env(self, obstacle_name, obstacle_info, target_name, target_info): self.obs_r = [] self.obs_pos = [] self.obs_ori = [] self.obs_grid = [] self.tar_pos = [round(target_info[0][0][2], 2), round(-target_info[0][0][1], 2)] self.tar_grid = [int(round((target_info[0][0][2] - self.ws_zero[0]) * 100)), int(round((-target_info[0][0][1] - self.ws_zero[1]) * 100))] self.tar_r = round(target_info[0][2][1] * 0.5, 2) print "number of obstacles", len(obstacle_info) for i in range(len(obstacle_info)): self.obs_pos.append([round(obstacle_info[i][0][2], 2), round(-obstacle_info[i][0][1], 2)]) self.obs_grid.append([int(round((obstacle_info[i][0][2] - self.ws_zero[0])*100)), int(round((-obstacle_info[i][0][1] - self.ws_zero[1])*100))]) self.obs_r.append(round(obstacle_info[i][2][1] * 0.5, 2)) self.obs_ori.append([0.0, 0.0, 0.0]) self.object_z = obstacle_info[0][0][0] print "obstacles", self.obs_pos print "target_vfh", self.tar_pos print "target_rviz", target_info[0] print "robot pos", self.rob_pos tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) self.ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop()
def get_c_ore(self, in_can_info): # print "input", in_can_info # print in_can_info[0].pos t_c_order = [] for ci in range(len(in_can_info)): tm_tar_pos = in_can_info[ci].pos tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) # tm_obs_pos.extend(self.obs_wall) tm_ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(tm_ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(tm_ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() t_c_order.append(ore_order) return t_c_order
def get_c_ore(self, in_can_info): # print "input", in_can_info # print in_can_info[0].pos t_c_order = [] for ci in range(len(in_can_info)): tm_tar_pos = in_can_info[ci].pos tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) # tm_obs_pos.extend(self.obs_wall) tm_ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(tm_ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore( tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(tm_ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) while 1: tm_tar_pos = in_can_info[ci].pos tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) for i in ore_order: if i != 'T': tm_obs_pos[i] = [3.0, 3.0] tmp_order = NG_ore( tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order if tmp_order[0] != 'T': print "tricky environment for c_ore so extend", ore_order ore_order.pop() print "delete target", ore_order ore_order.extend(tmp_order) print "to", ore_order else: if tmp_order[0] == 'T': print "ok", ore_order break ore_order.pop() if len(tm_obs_pos) in ore_order: print "\n\nThere is target!! warning!!!\n\n" t_c_order = [] else: t_c_order.append(ore_order) return t_c_order
def get_can_info(self, in_can_info, in_obs_pos, in_obs_re_pos, in_ore_order, in_tar_pos): tmp_can_info = [] for i in range(len(in_ore_order)): tmp_can_info.append(copy.deepcopy(in_can_info)) tmp_obs_pos = copy.deepcopy(in_obs_pos) tmp_obs_re_pos = copy.deepcopy(in_obs_re_pos) tmp_ore_order = copy.deepcopy(in_ore_order) tmp_tar_pos = copy.deepcopy(in_tar_pos) # print("\nCheck if candidate blocks the target") for step_i in range(len(tmp_ore_order)): for i in range(len(tmp_can_info[step_i])): vfh_obs_pos = copy.deepcopy(tmp_obs_re_pos) vfh_obs_pos.append(tmp_can_info[step_i][i].pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(tmp_tar_pos) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.obs_r, self.tar_r) if vfh[3] == 0: tmp_can_info[step_i][i].BT = 1 # BT == 1 : The candidate blocks the target. else: # BT == 0 : The candidate does not block the target. tmp_can_info[step_i][i].BT = 0 # print("\nCheck if the candidate is accessible.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) for si in range(step_i+1): # print "\nstep", si, "\nbefore", vfh_obs_pos vfh_obs_pos.remove(tmp_obs_pos[tmp_ore_order[si]]) # print "after", vfh_obs_pos vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. else: # tmp_can_info[step_i][i].A = 1 # print("\nCheck the candidate ORC.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. tm_tar_pos = copy.deepcopy(vfh_tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(tmp_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for obs_ori_i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() # The last order is always the target so we need to pop the last element. tmp_can_info[step_i][i].ORC = ore_order else: # tmp_can_info[step_i][i].A = 1 return tmp_can_info
def get_env_case1(self): while 1: obs_r = [0.035, 0.035, 0.035, 0.035, 0.035, 0.035, 0.035, 0.035] tar_r = 0.035 min_ore = 2 self.obs_grid = [] grid_tmp = copy.deepcopy(self.grid_act) self.obs_r = obs_r self.tar_r = tar_r self.obs_grid = [[10, 70], [10, 50], [10, 10], [20, 70], [20, 50], [19, 31], [24, 25], [20, 10]] self.tar_grid = [30, 30] # self.grid_ori = copy.deepcopy(self.grid_act) # for i in range(len(self.obs_r)): # print "obs", i, round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2) # self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) # self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target # break # for ri in self.obs_r: # grid_tmp, obs_center_tmp = CUF.place_circle_object_ig(grid_tmp, ri, 2) # self.obs_grid.append(obs_center_tmp) # grid_tmp, tar_tmp = CUF.place_circle_object_ig(grid_tmp, self.tar_r, 4) # self.tar_grid = copy.deepcopy(tar_tmp) self.obs_pos = [] for i in self.obs_grid: xi, yi = i self.obs_pos.append([round(xi * self.GRID_SIZE + self.ws_zero[0], 2), round(yi * self.GRID_SIZE + self.ws_zero[1], 2)]) self.tar_pos = [round(self.tar_grid[0] * self.GRID_SIZE + self.ws_zero[0], 2), round(self.tar_grid[1] * self.GRID_SIZE + self.ws_zero[1], 2)] # target object! self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) if len(ore_order) > min_ore: # print"environment setting OK" # print"target", self.tar_pos # print"obstacles", self.obs_pos # print"remove", ore_order, "th obstacles" tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) # self.ore_order = ore_order self.ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_ore_pos.append(self.obs_pos[i]) for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) # tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) if len(tmp_order) == 1: print"environment setting OK" break ''' # with out additional rearrangement ''' if len(self.ore_order) > min_ore: self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop()
class EnvInfo: def __init__(self, rob_pos, ws_width, ws_depth, ws_cen, grid_size, wall_r): self.rob_pos = rob_pos self.GRID_SIZE = grid_size self.ws_w = ws_width self.ws_d = ws_depth self.ws_cen = ws_cen self.ws_zero = [round(self.ws_cen[0] - ws_width * self.GRID_SIZE * 0.5, 2), round(self.ws_cen[1] - ws_depth * self.GRID_SIZE * 0.5, 2)] self.obs_wall = self.get_obs_wall(OBJ_R=wall_r) self.order_error_flag = 1 self.d_max = 2.0 self.eta = 45 grid_act = np.zeros([ws_width, ws_depth]) self.grid_act = CUF.mark_edge_grid(grid_act) def get_env(self, obs_r, tar_r, min_ore): while 1: self.obs_grid = [] grid_tmp = copy.deepcopy(self.grid_act) self.obs_r = obs_r self.tar_r = tar_r for ri in self.obs_r: grid_tmp, obs_center_tmp = CUF.place_circle_object_ig(grid_tmp, ri, 2) self.obs_grid.append(obs_center_tmp) grid_tmp, tar_tmp = CUF.place_circle_object_ig(grid_tmp, self.tar_r, 4) self.tar_grid = copy.deepcopy(tar_tmp) self.obs_pos = [] for i in self.obs_grid: xi, yi = i self.obs_pos.append([round(xi * self.GRID_SIZE + self.ws_zero[0], 2), round(yi * self.GRID_SIZE + self.ws_zero[1], 2)]) self.tar_pos = [round(self.tar_grid[0] * self.GRID_SIZE + self.ws_zero[0], 2), round(self.tar_grid[1] * self.GRID_SIZE + self.ws_zero[1], 2)] # target object! self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # TM_noplot() if len(ore_order) > min_ore: # print"before rearrangemet: ", ore_order tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) # ob = len(tm_obs_pos) self.ore_order = ore_order # self.ore_order = TM_plot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order if tmp_order[0] == 'T': # print"environment setting OK" break # else: # print "SHORT...min:", min_ore, "ours:", len(ore_order) ''' with out additional rearrangement ''' if len(self.ore_order) > min_ore: self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() def get_env_case1(self): while 1: obs_r = [0.035, 0.035, 0.035, 0.035, 0.035, 0.035, 0.035, 0.035] tar_r = 0.035 min_ore = 2 self.obs_grid = [] grid_tmp = copy.deepcopy(self.grid_act) self.obs_r = obs_r self.tar_r = tar_r self.obs_grid = [[10, 70], [10, 50], [10, 10], [20, 70], [20, 50], [19, 31], [24, 25], [20, 10]] self.tar_grid = [30, 30] # self.grid_ori = copy.deepcopy(self.grid_act) # for i in range(len(self.obs_r)): # print "obs", i, round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2) # self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) # self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target # break # for ri in self.obs_r: # grid_tmp, obs_center_tmp = CUF.place_circle_object_ig(grid_tmp, ri, 2) # self.obs_grid.append(obs_center_tmp) # grid_tmp, tar_tmp = CUF.place_circle_object_ig(grid_tmp, self.tar_r, 4) # self.tar_grid = copy.deepcopy(tar_tmp) self.obs_pos = [] for i in self.obs_grid: xi, yi = i self.obs_pos.append([round(xi * self.GRID_SIZE + self.ws_zero[0], 2), round(yi * self.GRID_SIZE + self.ws_zero[1], 2)]) self.tar_pos = [round(self.tar_grid[0] * self.GRID_SIZE + self.ws_zero[0], 2), round(self.tar_grid[1] * self.GRID_SIZE + self.ws_zero[1], 2)] # target object! self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) if len(ore_order) > min_ore: # print"environment setting OK" # print"target", self.tar_pos # print"obstacles", self.obs_pos # print"remove", ore_order, "th obstacles" tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) # self.ore_order = ore_order self.ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_ore_pos.append(self.obs_pos[i]) for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) # tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) if len(tmp_order) == 1: print"environment setting OK" break ''' # with out additional rearrangement ''' if len(self.ore_order) > min_ore: self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() def get_max_can_case1(self): # [19, 35], [24, 25] self.can_grid = [[30, 70], [30, 50], [14, 25], [30, 10]] self.can_pos = [] circle_r = max(self.ore_r)+0.005 for i in self.can_grid: xi, yi = i self.can_pos.append([self.ws_zero[0] + xi * self.GRID_SIZE, self.ws_zero[1] + yi * self.GRID_SIZE]) self.grid_max_can = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_max_can = CUF.obstacle_circle(self.grid_max_can, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_max_can = CUF.obstacle_circle(self.grid_max_can, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target for i in range(len(self.can_grid)): self.grid_max_can = CUF.obstacle_circle(self.grid_max_can, [self.can_grid[i][0], self.can_grid[i][1], circle_r], 3) def update_env(self, in_obs_pos, in_obs_grid): self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(in_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) self.ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # self.ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order # if tmp_order[0] == 'T': # print "environment setting OK" self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(in_obs_grid) self.obs_re_pos = copy.deepcopy(in_obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() # else: # self.order_error_flag = 0 def get_max_can(self, input_grid, bt_num, trial_num, ): bt_circle = [] circle_r = max(self.ore_r)+0.03 for bt in range(bt_num): can_grid = [] grid_can = copy.deepcopy(input_grid) # get original scene from the grid_set empt_grid, occu_grid = CUF.getEmpOcc(grid_can) for i in range(trial_num): pick_cen = np.random.randint(0, len(empt_grid)) check_sum = 0 for oc in range(len(occu_grid)): d_w = empt_grid[pick_cen][0] - occu_grid[oc][0] d_d = empt_grid[pick_cen][1] - occu_grid[oc][1] d_c = (d_w * d_w + d_d * d_d) ** 0.5 * self.GRID_SIZE if d_c <= circle_r: check_sum = 1 if check_sum == 0: can_grid.append(empt_grid[pick_cen]) for em in range(len(empt_grid)): d_w = empt_grid[pick_cen][0] - empt_grid[em][0] d_d = empt_grid[pick_cen][1] - empt_grid[em][1] d_c = (d_w * d_w + d_d * d_d) ** 0.5 * self.GRID_SIZE if d_c <= circle_r: grid_can[empt_grid[em][0]][empt_grid[em][1]] = 3 grid_can[empt_grid[pick_cen][0]][empt_grid[pick_cen][1]] = 3 occu_grid.append([empt_grid[em][0], empt_grid[em][1]]) bt_circle.append([can_grid, grid_can]) max_cir_num = [] for i in range(len(bt_circle)): max_cir_num.append([len(bt_circle[i][0])]) # print(max_cir_num.index(max(max_cir_num))) max_trial = max_cir_num.index(max(max_cir_num)) self.grid_max_can = copy.deepcopy(bt_circle[max_trial][1]) self.can_grid = bt_circle[max_trial][0] self.can_pos = [] for i in self.can_grid: xi, yi = i self.can_pos.append([self.ws_zero[0] + xi * self.GRID_SIZE, self.ws_zero[1] + yi * self.GRID_SIZE]) def get_obs_wall(self, OBJ_R): ws_side = [] ws_side.append( [self.ws_cen[0] - self.ws_w * self.GRID_SIZE * 0.5, self.ws_cen[1] - self.ws_d * self.GRID_SIZE * 0.5 - OBJ_R]) # left low point ws_side.append([self.ws_cen[0] + self.ws_w * self.GRID_SIZE * 0.5 + OBJ_R, self.ws_cen[1] - self.ws_d * self.GRID_SIZE * 0.5 - OBJ_R]) # right low point ws_side.append([self.ws_cen[0] + self.ws_w * self.GRID_SIZE * 0.5 + OBJ_R, self.ws_cen[1] + self.ws_d * self.GRID_SIZE * 0.5 + OBJ_R]) # right high point ws_side.append( [self.ws_cen[0] - self.ws_w * self.GRID_SIZE * 0.5, self.ws_cen[1] + self.ws_d * self.GRID_SIZE * 0.5 + OBJ_R]) # left high point obs_wall = [] obs_wall.extend(CUF.linspace2D(ws_side[0], ws_side[1], round(self.ws_w * self.GRID_SIZE / OBJ_R))) obs_wall.extend(CUF.linspace2D(ws_side[1], ws_side[2], round(self.ws_d * self.GRID_SIZE / OBJ_R))) obs_wall.extend(CUF.linspace2D(ws_side[2], ws_side[3], round(self.ws_w * self.GRID_SIZE / OBJ_R))) return obs_wall def get_can_info(self, in_can_info, in_obs_pos, in_obs_re_pos, in_ore_order, in_tar_pos): tmp_can_info = [] for i in range(len(in_ore_order)): tmp_can_info.append(copy.deepcopy(in_can_info)) tmp_obs_pos = copy.deepcopy(in_obs_pos) tmp_obs_re_pos = copy.deepcopy(in_obs_re_pos) tmp_ore_order = copy.deepcopy(in_ore_order) tmp_tar_pos = copy.deepcopy(in_tar_pos) # print("\nCheck if candidate blocks the target") for step_i in range(len(tmp_ore_order)): for i in range(len(tmp_can_info[step_i])): vfh_obs_pos = copy.deepcopy(tmp_obs_re_pos) vfh_obs_pos.append(tmp_can_info[step_i][i].pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(tmp_tar_pos) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.obs_r, self.tar_r) if vfh[3] == 0: tmp_can_info[step_i][i].BT = 1 # BT == 1 : The candidate blocks the target. else: # BT == 0 : The candidate does not block the target. tmp_can_info[step_i][i].BT = 0 # print("\nCheck if the candidate is accessible.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) for si in range(step_i+1): # print "\nstep", si, "\nbefore", vfh_obs_pos vfh_obs_pos.remove(tmp_obs_pos[tmp_ore_order[si]]) # print "after", vfh_obs_pos vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. else: # tmp_can_info[step_i][i].A = 1 # print("\nCheck the candidate ORC.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. tm_tar_pos = copy.deepcopy(vfh_tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(tmp_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for obs_ori_i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() # The last order is always the target so we need to pop the last element. tmp_can_info[step_i][i].ORC = ore_order else: # tmp_can_info[step_i][i].A = 1 return tmp_can_info def get_can_A(self, in_can_info, in_obs_pos, in_tar_pos): tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck if the candidate is accessible.") for ci in range(len(tmp_can_info)): vfh_tar_pos = copy.deepcopy(tmp_can_info[ci].pos) vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.append(copy.deepcopy(in_tar_pos)) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[ci].A = 0 # A == 0 : The candidate is not accessible. else: # tmp_can_info[ci].A = 1 return tmp_can_info def get_can_BT(self, in_can_info, in_obs_pos, in_tar_pos): tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck if candidate blocks the target") for ci in range(len(tmp_can_info)): vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.append(tmp_can_info[ci].pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(in_tar_pos) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: tmp_can_info[ci].BT = 1 # BT == 1 : The candidate blocks the target. # else: # BT == 0 : The candidate does not block the target. # tmp_can_info[ci].BT = 0 # print "can", ci, "bt=0" return tmp_can_info def get_cf(self, in_can_info): tmp_cf = [] tmp_cf_index = [] tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck the candidate ORC.") for ci in range(len(tmp_can_info)): # print "\ncan ", ci, "th has A, BT :", tmp_can_info[ci].A, tmp_can_info[ci].BT if tmp_can_info[ci].A == 1 and tmp_can_info[ci].BT == 0: tmp_cf.append(tmp_can_info[ci]) tmp_cf_index.append(ci) return tmp_cf, tmp_cf_index def get_cf_b(self, in_cf, in_obs_pos): tmp_cf = copy.deepcopy(in_cf) tmp_obs_pos = copy.deepcopy(in_obs_pos) tmp_b = [] for cb in range(len(tmp_cf)): # cb: The candidate that will check the b value b = 0 for ci in range(len(tmp_cf)): # ci: Other candidates for checking the b value if cb != ci: # print "\ntar", tmp_cf_pos[ci] # print "obs", tmp_obs_pos vfh_tar_pos = copy.deepcopy(tmp_cf[ci].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) vfh_obs_pos.append(tmp_cf[cb].pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: b = b + 1 tmp_b.append(b) return tmp_b def get_cp(self, in_can_info): tmp_cp = [] tmp_cp_index = [] tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck the candidate ORC.") for ci in range(len(tmp_can_info)): # print "\ncan ", ci, "th has A, BT :", tmp_can_info[ci].A, tmp_can_info[ci].BT if tmp_can_info[ci].A == 0: tmp_cp.append(tmp_can_info[ci]) tmp_cp_index.append(ci) return tmp_cp, tmp_cp_index # def get_can_A(self, in_can_info, in_obs_pos, in_tar_pos): # tmp_can_info = copy.deepcopy(in_can_info) # # print("\nCheck if the candidate is accessible.") # for ci in range(len(tmp_can_info)): # vfh_tar_pos = copy.deepcopy(tmp_can_info[ci].pos) # vfh_obs_pos = copy.deepcopy(in_obs_pos) # vfh_obs_pos.append(copy.deepcopy(in_tar_pos)) # vfh_obs_pos.extend(self.obs_wall) # ob = len(vfh_obs_pos) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # if vfh[3] == 0: # A == 1 : The candidate is accessible. # tmp_can_info[ci].A = 0 # A == 0 : The candidate is not accessible. # # print "c:", ci, ".A = 0" # else: # # tmp_can_info[ci].A = 1 # # print "c:", ci, ".A = 1" # return tmp_can_info # # def get_can_BT(self, in_can_info, in_obs_pos, in_tar_pos): # tmp_can_info = copy.deepcopy(in_can_info) # # print("\nCheck if candidate blocks the target") # for ci in range(len(tmp_can_info)): # vfh_obs_pos = copy.deepcopy(in_obs_pos) # vfh_obs_pos.append(tmp_can_info[ci].pos) # vfh_obs_pos.extend(self.obs_wall) # vfh_tar_pos = copy.deepcopy(in_tar_pos) # ob = len(vfh_obs_pos) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # if vfh[3] == 0: # tmp_can_info[ci].BT = 1 # BT == 1 : The candidate blocks the target. # # print "c:", ci, ".BT = 1" # # else: # BT == 0 : The candidate does not block the target. # # tmp_can_info[ci].BT = 0 # # print "c:", ci, ".BT = 0" # # return tmp_can_info def get_c_ore(self, in_can_info): # print "input", in_can_info # print in_can_info[0].pos t_c_order = [] for ci in range(len(in_can_info)): tm_tar_pos = in_can_info[ci].pos tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) # tm_obs_pos.extend(self.obs_wall) tm_ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(tm_ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(tm_ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() t_c_order.append(ore_order) return t_c_order
class EnvInfo: def __init__(self, rob_pos, ws_width, ws_depth, ws_cen, grid_size, wall_r): self.rob_pos = rob_pos self.GRID_SIZE = grid_size self.ws_w = ws_width self.ws_d = ws_depth self.ws_cen = ws_cen self.ws_zero = [round(self.ws_cen[0] - ws_width * self.GRID_SIZE * 0.5, 2), round(self.ws_cen[1] - ws_depth * self.GRID_SIZE * 0.5, 2)] self.obs_wall = self.get_obs_wall(OBJ_R=wall_r) self.jaco_home_pos = [0.020, -0.0046, 0.673] self.jaco_home_ori = [-0.1869, 0.125044, 0.69344, 0.6845144] # goal_pose = [-0.010, -0.005, 0.967] # goal_orientation = [-0.1869, 0.125044, 0.69344, 0.6845144] self.order_error_flag = 1 self.d_max = 2.0 self.eta = 45 grid_act = np.zeros([ws_width, ws_depth]) self.grid_act = CUF.mark_edge_grid(grid_act) def set_env(self, obstacle_name, obstacle_info, target_name, target_info): self.obs_r = [] self.obs_pos = [] self.obs_ori = [] self.obs_grid = [] self.tar_pos = [round(target_info[0][0][2], 2), round(-target_info[0][0][1], 2)] self.tar_grid = [int(round((target_info[0][0][2] - self.ws_zero[0]) * 100)), int(round((-target_info[0][0][1] - self.ws_zero[1]) * 100))] self.tar_r = round(target_info[0][2][1] * 0.5, 2) print "number of obstacles", len(obstacle_info) for i in range(len(obstacle_info)): self.obs_pos.append([round(obstacle_info[i][0][2], 2), round(-obstacle_info[i][0][1], 2)]) self.obs_grid.append([int(round((obstacle_info[i][0][2] - self.ws_zero[0])*100)), int(round((-obstacle_info[i][0][1] - self.ws_zero[1])*100))]) self.obs_r.append(round(obstacle_info[i][2][1] * 0.5, 2)) self.obs_ori.append([0.0, 0.0, 0.0]) self.object_z = obstacle_info[0][0][0] print "obstacles", self.obs_pos print "target_vfh", self.tar_pos print "target_rviz", target_info[0] print "robot pos", self.rob_pos tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) self.ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() def get_env(self, obs_r, tar_r, min_ore): while 1: self.obs_grid = [] grid_tmp = copy.deepcopy(self.grid_act) self.obs_r = obs_r self.tar_r = tar_r for ri in self.obs_r: grid_tmp, obs_center_tmp = CUF.place_circle_object_ig(grid_tmp, ri, 2) self.obs_grid.append(obs_center_tmp) grid_tmp, tar_tmp = CUF.place_circle_object_ig(grid_tmp, self.tar_r, 4) self.tar_grid = copy.deepcopy(tar_tmp) self.obs_pos = [] for i in self.obs_grid: xi, yi = i self.obs_pos.append([round(xi * self.GRID_SIZE + self.ws_zero[0], 2), round(yi * self.GRID_SIZE + self.ws_zero[1], 2)]) self.tar_pos = [round(self.tar_grid[0] * self.GRID_SIZE + self.ws_zero[0], 2), round(self.tar_grid[1] * self.GRID_SIZE + self.ws_zero[1], 2)] # target object! self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # TM_noplot() if len(ore_order) > min_ore: # print"before rearrangemet: ", ore_order tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) # tm_obs_pos.extend(self.obs_wall) # ob = len(tm_obs_pos) self.ore_order = ore_order # self.ore_order = TM_plot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order if tmp_order[0] == 'T': # print"environment setting OK" break # else: # print "SHORT...min:", min_ore, "ours:", len(ore_order) ''' with out additional rearrangement ''' if len(self.ore_order) > min_ore: self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(self.obs_grid) self.obs_re_pos = copy.deepcopy(self.obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() def update_env(self, in_obs_pos, in_obs_grid): self.d_max = 2.0 tm_tar_pos = copy.deepcopy(self.tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(in_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) self.ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # self.ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall, self.rob_pos, self.rob_pos, self.d_max) # check if the obstacles are rearranged then target is reachable tm_tar_pos = copy.deepcopy(self.tar_pos) tm_obs_pos = copy.deepcopy(self.obs_pos) tm_ore_pos = [] for i in self.ore_order: if i != 'T': tm_obs_pos.remove(self.obs_pos[i]) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tmp_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # tmp_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) # print "after removing:", tmp_order # if tmp_order[0] == 'T': # print "environment setting OK" self.ore_grid = [] self.ore_pos = [] self.ore_r = [] self.obs_re_grid = copy.deepcopy(in_obs_grid) self.obs_re_pos = copy.deepcopy(in_obs_pos) self.obs_re_r = copy.deepcopy(self.obs_r) for i in self.ore_order: if i != 'T': self.ore_grid.append(self.obs_re_grid[i]) self.ore_pos.append(self.obs_re_pos[i]) self.ore_r.append(self.obs_re_r[i]) for i in self.ore_order: if i != 'T': self.obs_re_grid.remove(self.obs_grid[i]) self.obs_re_pos.remove(self.obs_pos[i]) self.obs_re_r.remove(self.obs_r[i]) self.grid_ori = copy.deepcopy(self.grid_act) for i in range(len(self.obs_r)): self.grid_ori = CUF.obstacle_circle(self.grid_ori, [round(self.obs_grid[i][0], 2), round(self.obs_grid[i][1], 2), self.obs_r[i]], 2) self.grid_ori = CUF.obstacle_circle(self.grid_ori, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.grid_del = copy.deepcopy(self.grid_act) for i in range(len(self.obs_re_r)): self.grid_del = CUF.obstacle_circle(self.grid_del, [round(self.obs_re_grid[i][0], 2), round(self.obs_re_grid[i][1], 2), self.obs_re_r[i]], 2) self.grid_del = CUF.obstacle_circle(self.grid_del, [self.tar_grid[0], self.tar_grid[1], self.tar_r], 4) # target self.ore_order.pop() # else: # self.order_error_flag = 0 def get_max_can(self, input_grid, bt_num, trial_num, ): bt_circle = [] can_grid = [] # circle_r = max(self.ore_r)+0.02 circle_r = max(self.ore_r)+0.03 for bt in range(bt_num): grid_can = copy.deepcopy(input_grid) # get original scene from the grid_set empt_grid, occu_grid = CUF.getEmpOcc(grid_can) for i in range(trial_num): pick_cen = np.random.randint(0, len(empt_grid)) check_sum = 0 for oc in range(len(occu_grid)): d_w = empt_grid[pick_cen][0] - occu_grid[oc][0] d_d = empt_grid[pick_cen][1] - occu_grid[oc][1] d_c = (d_w * d_w + d_d * d_d) ** 0.5 * self.GRID_SIZE if d_c <= circle_r: check_sum = 1 if check_sum == 0: can_grid.append(empt_grid[pick_cen]) for em in range(len(empt_grid)): d_w = empt_grid[pick_cen][0] - empt_grid[em][0] d_d = empt_grid[pick_cen][1] - empt_grid[em][1] d_c = (d_w * d_w + d_d * d_d) ** 0.5 * self.GRID_SIZE if d_c <= circle_r: grid_can[empt_grid[em][0]][empt_grid[em][1]] = 3 grid_can[empt_grid[pick_cen][0]][empt_grid[pick_cen][1]] = 3 occu_grid.append([empt_grid[em][0], empt_grid[em][1]]) bt_circle.append([can_grid, grid_can]) max_cir_num = [] for i in range(len(bt_circle)): max_cir_num.append([len(bt_circle[i][0])]) # print(max_cir_num.index(max(max_cir_num))) max_trial = max_cir_num.index(max(max_cir_num)) self.grid_max_can = copy.deepcopy(bt_circle[max_trial][1]) self.can_grid = bt_circle[max_trial][0] self.can_pos = [] for i in self.can_grid: xi, yi = i self.can_pos.append([self.ws_zero[0] + xi * self.GRID_SIZE, self.ws_zero[1] + yi * self.GRID_SIZE]) def get_obs_wall(self, OBJ_R): ws_side = [] ws_side.append( [self.ws_cen[0] - self.ws_w * self.GRID_SIZE * 0.5, self.ws_cen[1] - self.ws_d * self.GRID_SIZE * 0.5 - OBJ_R]) # left low point ws_side.append([self.ws_cen[0] + self.ws_w * self.GRID_SIZE * 0.5 + OBJ_R, self.ws_cen[1] - self.ws_d * self.GRID_SIZE * 0.5 - OBJ_R]) # right low point ws_side.append([self.ws_cen[0] + self.ws_w * self.GRID_SIZE * 0.5 + OBJ_R, self.ws_cen[1] + self.ws_d * self.GRID_SIZE * 0.5 + OBJ_R]) # right high point ws_side.append( [self.ws_cen[0] - self.ws_w * self.GRID_SIZE * 0.5, self.ws_cen[1] + self.ws_d * self.GRID_SIZE * 0.5 + OBJ_R]) # left high point obs_wall = [] obs_wall.extend(CUF.linspace2D(ws_side[0], ws_side[1], round(self.ws_w * self.GRID_SIZE / OBJ_R))) obs_wall.extend(CUF.linspace2D(ws_side[1], ws_side[2], round(self.ws_d * self.GRID_SIZE / OBJ_R))) obs_wall.extend(CUF.linspace2D(ws_side[2], ws_side[3], round(self.ws_w * self.GRID_SIZE / OBJ_R))) return obs_wall def get_can_info(self, in_can_info, in_obs_pos, in_obs_re_pos, in_ore_order, in_tar_pos): tmp_can_info = [] for i in range(len(in_ore_order)): tmp_can_info.append(copy.deepcopy(in_can_info)) tmp_obs_pos = copy.deepcopy(in_obs_pos) tmp_obs_re_pos = copy.deepcopy(in_obs_re_pos) tmp_ore_order = copy.deepcopy(in_ore_order) tmp_tar_pos = copy.deepcopy(in_tar_pos) # print("\nCheck if candidate blocks the target") for step_i in range(len(tmp_ore_order)): for i in range(len(tmp_can_info[step_i])): vfh_obs_pos = copy.deepcopy(tmp_obs_re_pos) vfh_obs_pos.append(tmp_can_info[step_i][i].pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(tmp_tar_pos) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.obs_r, self.tar_r) if vfh[3] == 0: tmp_can_info[step_i][i].BT = 1 # BT == 1 : The candidate blocks the target. else: # BT == 0 : The candidate does not block the target. tmp_can_info[step_i][i].BT = 0 # print("\nCheck if the candidate is accessible.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) for si in range(step_i+1): # print "\nstep", si, "\nbefore", vfh_obs_pos vfh_obs_pos.remove(tmp_obs_pos[tmp_ore_order[si]]) # print "after", vfh_obs_pos vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. else: # tmp_can_info[step_i][i].A = 1 # print("\nCheck the candidate ORC.") for i in range(len(tmp_can_info[step_i])): vfh_tar_pos = copy.deepcopy(tmp_can_info[step_i][i].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) vfh_obs_pos.append(tmp_tar_pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. tmp_can_info[step_i][i].A = 0 # A == 0 : The candidate is not accessible. tm_tar_pos = copy.deepcopy(vfh_tar_pos) tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(tmp_obs_pos) # tm_obs_pos.extend(self.obs_wall) ob = len(tm_obs_pos) tm_obs_ori = [] for obs_ori_i in range(ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() # The last order is always the target so we need to pop the last element. tmp_can_info[step_i][i].ORC = ore_order else: # tmp_can_info[step_i][i].A = 1 return tmp_can_info def get_can_A(self, in_can_info, in_obs_pos, in_tar_pos): tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck if the candidate is accessible.") for ci in range(len(tmp_can_info)): vfh_tar_pos = copy.deepcopy(tmp_can_info[ci].pos) vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.append(copy.deepcopy(in_tar_pos)) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: # A == 1 : The candidate is accessible. print ci, "A = 0 (vfh)" tmp_can_info[ci].A = 0 # A == 0 : The candidate is not accessible. else: # print ci, "A = 1 (vfh)" print "\nangle:", vfh[-1] xi, yi = tmp_can_info[ci].pos[0], tmp_can_info[ci].pos[1] CLF.add_box_client('can_check', [self.object_z, -yi, xi], [-0.707, 0.0, -0.707, 0.0], [0.06, 0.06, 0.2], 'pink') planner_name = 'RRTConnect' # planner_name = 'BiTRRT' n_attempt = 10 c_time = 0.5 n_repeat = 5 start_state = moveit_msgs.msg.RobotState() joint_state = sensor_msgs.msg.JointState() joint_state.header = std_msgs.msg.Header() joint_state.name = ['j2n6s300_joint_1', 'j2n6s300_joint_2', 'j2n6s300_joint_3', 'j2n6s300_joint_4', 'j2n6s300_joint_5', 'j2n6s300_joint_6'] joint_state.position = [3.1415927410125732, 4.537856101989746, 5.93411922454834, -0.6108652353286743, 1.7453292608261108, -0.5235987901687622] start_state.joint_state = joint_state # goal_pose: # goal_orientation: # goal_pose = [self.object_z, -yi, xi - 0.05] # goal_orientation = [-0.00145713772037, -0.998970756926, 0.0364956710831, 0.0268955302573] z = self.object_z goal_pose = [z + 0.02, -yi, xi] # Set the grasp pose: substract 17cm from the z value of the object centroid goal_pitches = [] goal_pitch = np.deg2rad(vfh[-1]) # goal_pitch = vfh[-1] + math.pi/2 goal_pitches.append(goal_pitch) # approaching_angle: vfh[-1] from the input for i in range(1): goal_pitches.append(goal_pitch + (i + 1) * (math.pi / 36)) goal_pitches.append(goal_pitch - (i + 1) * (math.pi / 36)) # Get the grasp orientation (currently the front direction) goal_orientations = [] for i in goal_pitches: goal_orientations.append(quaternion_from_euler(-i, math.radians(-5.0), math.radians(90.0), axes='rxyz')) l = 0.001 goal_poses = [] for i in goal_pitches: dx = math.sin(i - math.pi) * l dy = math.cos(i - math.pi) * l goal_poses.append([z + 0.07, -yi + dx, xi + dy]) # goal_poses.append([z + 0.07, goal_pose[1] + dx, goal_pose[2] - dy]) # plt.figure() # for i in range(len(in_obs_pos)): # plt.scatter(in_obs_pos[i][0], in_obs_pos[i][1], s=200, c='red') # plt.scatter(xi, yi, s=200, c='pink') # plt.scatter(xi + dy, yi + dx, s=200, c='blue') # # plt.figure() # for i in range(len(in_obs_pos)): # plt.scatter(in_obs_pos[i][0], in_obs_pos[i][1], s=200, c='red') # plt.scatter(xi, yi, s=200, c='pink') # plt.scatter(xi + dy, yi - dx, s=200, c='blue') # # plt.show() feasibility1 = 0 i = 0 while not feasibility1 and i < len(goal_pitches): CLF.add_box_client('can_bottom_x', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.1, 0.005, 0.005], 'red') CLF.add_box_client('can_bottom_y', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.1, 0.005], 'blue') CLF.add_box_client('can_bottom_z', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.005, 0.1], 'green') [feasibility1, trajectory1] = CLF.feasible_check_obj_joint_client('arm', 'gripper', start_state, goal_poses[i], goal_orientations[i], [], planner_name, n_attempt, c_time, n_repeat) i = i + 1 time.sleep(1) CLF.del_box_client('can_bottom_x') CLF.del_box_client('can_bottom_y') CLF.del_box_client('can_bottom_z') if feasibility1 == 0: # A == 1 : The candidate is accessible. print ci, "A = 0 (MP)" tmp_can_info[ci].A = 0 # A == 0 : The candidate is not accessible. else: # print ci, "A = 1 (MP)" tmp_can_info[ci].A = 1 CLF.del_box_client('can_check') tmp_can_info[ci].A = 1 return tmp_can_info def get_can_BT(self, in_can_info, in_obs_pos, in_tar_pos): tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck if candidate blocks the target") for ci in range(len(tmp_can_info)): if tmp_can_info[ci].A == 1: vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.append(tmp_can_info[ci].pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(in_tar_pos) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: print ci, "BT = 1 (vfh)" tmp_can_info[ci].BT = 1 # BT == 1 : The candidate blocks the target. else: # BT == 0 : The candidate does not block the target. print ci, "BT = 0 (vfh)" print "\nangle:", vfh[-1] xi, yi = tmp_can_info[ci].pos[0], tmp_can_info[ci].pos[1] CLF.add_box_client('can_check', [self.object_z, -yi, xi], [-0.707, 0.0, -0.707, 0.0], [0.06, 0.06, 0.2], 'pink') planner_name = 'RRTConnect' # planner_name = 'BiTRRT' n_attempt = 10 c_time = 0.5 n_repeat = 5 start_state = moveit_msgs.msg.RobotState() joint_state = sensor_msgs.msg.JointState() joint_state.header = std_msgs.msg.Header() joint_state.name = ['j2n6s300_joint_1', 'j2n6s300_joint_2', 'j2n6s300_joint_3', 'j2n6s300_joint_4', 'j2n6s300_joint_5', 'j2n6s300_joint_6'] joint_state.position = [3.1415927410125732, 4.537856101989746, 5.93411922454834, -0.6108652353286743, 1.7453292608261108, -0.5235987901687622] start_state.joint_state = joint_state # goal_pose: # goal_orientation: # goal_pose = [self.object_z, -yi, xi - 0.05] # goal_orientation = [-0.00145713772037, -0.998970756926, 0.0364956710831, 0.0268955302573] z = self.object_z goal_pose = [z + 0.02, -yi, xi] # Set the grasp pose: substract 17cm from the z value of the object centroid goal_pitches = [] goal_pitch = np.deg2rad(vfh[-1]) # goal_pitch = vfh[-1] + math.pi/2 goal_pitches.append(goal_pitch) # approaching_angle: vfh[-1] from the input for i in range(1): goal_pitches.append(goal_pitch + (i + 1) * (math.pi / 36)) goal_pitches.append(goal_pitch - (i + 1) * (math.pi / 36)) # Get the grasp orientation (currently the front direction) goal_orientations = [] for i in goal_pitches: goal_orientations.append(quaternion_from_euler(-i, math.radians(-5.0), math.radians(90.0), axes='rxyz')) l = 0.001 goal_poses = [] for i in goal_pitches: dx = math.sin(i - math.pi) * l dy = math.cos(i - math.pi) * l goal_poses.append([z + 0.02, -in_tar_pos[1] + dx, in_tar_pos[0] + dy]) # goal_poses.append([z + 0.07, goal_pose[1] + dx, goal_pose[2] - dy]) # plt.figure() # for i in range(len(in_obs_pos)): # plt.scatter(in_obs_pos[i][0], in_obs_pos[i][1], s=200, c='red') # plt.scatter(xi, yi, s=200, c='pink') # plt.scatter(xi + dy, yi + dx, s=200, c='blue') # # plt.figure() # for i in range(len(in_obs_pos)): # plt.scatter(in_obs_pos[i][0], in_obs_pos[i][1], s=200, c='red') # plt.scatter(xi, yi, s=200, c='pink') # plt.scatter(xi + dy, yi - dx, s=200, c='blue') # # plt.show() feasibility1 = 0 i = 0 while not feasibility1 and i < len(goal_pitches): CLF.add_box_client('can_bottom_x', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.1, 0.005, 0.005], 'red') CLF.add_box_client('can_bottom_y', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.1, 0.005], 'blue') CLF.add_box_client('can_bottom_z', [goal_poses[i][0]-0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.005, 0.1], 'green') [feasibility1, trajectory1] = CLF.feasible_check_obj_joint_client('arm', 'gripper', start_state, goal_poses[i], goal_orientations[i], [], planner_name, n_attempt, c_time, n_repeat) i = i + 1 time.sleep(1) CLF.del_box_client('can_bottom_x') CLF.del_box_client('can_bottom_y') CLF.del_box_client('can_bottom_z') if feasibility1 == 0: # A == 1 : The candidate is accessible. print ci, "BT = 1 (MP)" tmp_can_info[ci].BT = 1 # A == 0 : The candidate is not accessible. else: # print ci, "BT = 0 (MP)" tmp_can_info[ci].BT = 0 CLF.del_box_client('can_check') else: print ci, "A = 0 => BT = no matter" return tmp_can_info def get_cf(self, in_can_info): tmp_cf = [] tmp_cf_index = [] tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck the candidate ORC.") for ci in range(len(tmp_can_info)): # print "\ncan ", ci, "th has A, BT :", tmp_can_info[ci].A, tmp_can_info[ci].BT if tmp_can_info[ci].A == 1 and tmp_can_info[ci].BT == 0: tmp_cf.append(tmp_can_info[ci]) tmp_cf_index.append(ci) return tmp_cf, tmp_cf_index def get_cf_b(self, in_cf, in_obs_pos): tmp_cf = copy.deepcopy(in_cf) tmp_obs_pos = copy.deepcopy(in_obs_pos) tmp_b = [] for cb in range(len(tmp_cf)): # cb: The candidate that will check the b value b = 0 for ci in range(len(tmp_cf)): # ci: Other candidates for checking the b value if cb != ci: # print "\ntar", tmp_cf_pos[ci] # print "obs", tmp_obs_pos vfh_tar_pos = copy.deepcopy(tmp_cf[ci].pos) vfh_obs_pos = copy.deepcopy(tmp_obs_pos) vfh_obs_pos.append(tmp_cf[cb].pos) vfh_obs_pos.extend(self.obs_wall) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 0: b = b + 1 tmp_b.append(b) return tmp_b def get_cp(self, in_can_info): tmp_cp = [] tmp_cp_index = [] tmp_can_info = copy.deepcopy(in_can_info) # print("\nCheck the candidate ORC.") for ci in range(len(tmp_can_info)): # print "\ncan ", ci, "th has A, BT :", tmp_can_info[ci].A, tmp_can_info[ci].BT if tmp_can_info[ci].A == 0: tmp_cp.append(tmp_can_info[ci]) tmp_cp_index.append(ci) return tmp_cp, tmp_cp_index # def get_can_A(self, in_can_info, in_obs_pos, in_tar_pos): # tmp_can_info = copy.deepcopy(in_can_info) # # print("\nCheck if the candidate is accessible.") # for ci in range(len(tmp_can_info)): # vfh_tar_pos = copy.deepcopy(tmp_can_info[ci].pos) # vfh_obs_pos = copy.deepcopy(in_obs_pos) # vfh_obs_pos.append(copy.deepcopy(in_tar_pos)) # vfh_obs_pos.extend(self.obs_wall) # ob = len(vfh_obs_pos) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # if vfh[3] == 0: # A == 1 : The candidate is accessible. # tmp_can_info[ci].A = 0 # A == 0 : The candidate is not accessible. # # print "c:", ci, ".A = 0" # else: # # tmp_can_info[ci].A = 1 # # print "c:", ci, ".A = 1" # return tmp_can_info # # def get_can_BT(self, in_can_info, in_obs_pos, in_tar_pos): # tmp_can_info = copy.deepcopy(in_can_info) # # print("\nCheck if candidate blocks the target") # for ci in range(len(tmp_can_info)): # vfh_obs_pos = copy.deepcopy(in_obs_pos) # vfh_obs_pos.append(tmp_can_info[ci].pos) # vfh_obs_pos.extend(self.obs_wall) # vfh_tar_pos = copy.deepcopy(in_tar_pos) # ob = len(vfh_obs_pos) # vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) # if vfh[3] == 0: # tmp_can_info[ci].BT = 1 # BT == 1 : The candidate blocks the target. # # print "c:", ci, ".BT = 1" # # else: # BT == 0 : The candidate does not block the target. # # tmp_can_info[ci].BT = 0 # # print "c:", ci, ".BT = 0" # # return tmp_can_info def get_c_ore(self, in_can_info): # print "input", in_can_info # print in_can_info[0].pos t_c_order = [] for ci in range(len(in_can_info)): tm_tar_pos = in_can_info[ci].pos tm_tar_ori = [0.0, 0.0, 0.0] tm_obs_pos = copy.deepcopy(self.obs_pos) tm_obs_pos.append(self.tar_pos) # tm_obs_pos.extend(self.obs_wall) tm_ob = len(tm_obs_pos) tm_obs_ori = [] for i in range(tm_ob): tm_obs_ori.append([0.0, 0.0, 0.0]) ore_order = NG_ore(tm_tar_pos, tm_obs_pos, self.tar_r, self.obs_r, self.rob_pos, self.ws_zero, [self.ws_w * self.GRID_SIZE, self.ws_d * self.GRID_SIZE]) # ore_order = TM_noplot(tm_ob, tm_tar_pos, tm_tar_ori, tm_obs_pos, tm_obs_ori, self.obs_wall,self.rob_pos, self.rob_pos, self.d_max) ore_order.pop() t_c_order.append(ore_order) return t_c_order def pick(self, in_obs_pos, pick_pose, current_joints): print "pick at:", pick_pose vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(pick_pose) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 1: print "\nangle:", vfh[-1] xi, yi = pick_pose[0], pick_pose[1] planner_name = 'RRTConnect' # planner_name = 'BiTRRT' n_attempt = 100 c_time = 3 n_repeat = 5 start_state = moveit_msgs.msg.RobotState() joint_state = sensor_msgs.msg.JointState() joint_state.header = std_msgs.msg.Header() joint_state.name = ['j2n6s300_joint_1', 'j2n6s300_joint_2', 'j2n6s300_joint_3', 'j2n6s300_joint_4', 'j2n6s300_joint_5', 'j2n6s300_joint_6'] joint_state.position = current_joints # joint_state.position = [3.1415927410125732, 4.537856101989746, 5.93411922454834, -0.6108652353286743, 1.7453292608261108, -0.5235987901687622] start_state.joint_state = joint_state # goal_pose: # goal_orientation: # goal_pose = [self.object_z, -yi, xi - 0.05] # goal_orientation = [-0.00145713772037, -0.998970756926, 0.0364956710831, 0.0268955302573] z = self.object_z # goal_pose = [z + 0.02, -yi, xi] # Set the grasp pose: substract 17cm from the z value of the object centroid goal_pitches = [] goal_pitch = np.deg2rad(vfh[-1]) # goal_pitch = vfh[-1] + math.pi/2 goal_pitches.append(goal_pitch) # approaching_angle: vfh[-1] from the input for i in range(5): goal_pitches.append(goal_pitch + (i + 1) * (math.pi / 36)) goal_pitches.append(goal_pitch - (i + 1) * (math.pi / 36)) # Get the grasp orientation (currently the front direction) goal_orientations = [] for i in goal_pitches: goal_orientations.append(quaternion_from_euler(-i, math.radians(-5.0), math.radians(90.0), axes='rxyz')) # l = 0.001 l = 0.17 goal_poses = [] for i in goal_pitches: dx = math.sin(i - math.pi) * l dy = math.cos(i - math.pi) * l goal_poses.append([z + 0.02, -pick_pose[1] + dx, pick_pose[0] + dy]) feasibility1 = 0 i = 0 while not feasibility1 and i < len(goal_pitches): # CLF.add_box_client('can_bottom_x', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.1, 0.005, 0.005], 'red') # CLF.add_box_client('can_bottom_y', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.1, 0.005], 'blue') # CLF.add_box_client('can_bottom_z', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.005, 0.1], 'green') [feasibility1, trajectory1] = CLF.move_goalpose_client('arm', 'gripper', start_state, goal_poses[i], goal_orientations[i], [], planner_name, n_attempt, c_time, n_repeat) i = i + 1 # time.sleep(1) # CLF.del_box_client('can_bottom_x') # CLF.del_box_client('can_bottom_y') # CLF.del_box_client('can_bottom_z') return [goal_poses[i], goal_orientations[i]] def place(self, in_obs_pos, place_pose, current_joints): print "place at:", place_pose vfh_obs_pos = copy.deepcopy(in_obs_pos) vfh_obs_pos.remove(place_pose) vfh_obs_pos.extend(self.obs_wall) vfh_tar_pos = copy.deepcopy(place_pose) ob = len(vfh_obs_pos) vfh = influence(ob, vfh_tar_pos, vfh_obs_pos, self.rob_pos, self.d_max, self.eta) if vfh[3] == 1: print "angle:", vfh[-1] xi, yi = place_pose[0], place_pose[1] planner_name = 'RRTConnect' # planner_name = 'BiTRRT' n_attempt = 100 c_time = 3 n_repeat = 5 start_state = moveit_msgs.msg.RobotState() joint_state = sensor_msgs.msg.JointState() joint_state.header = std_msgs.msg.Header() joint_state.name = ['j2n6s300_joint_1', 'j2n6s300_joint_2', 'j2n6s300_joint_3', 'j2n6s300_joint_4', 'j2n6s300_joint_5', 'j2n6s300_joint_6'] joint_state.position = current_joints # joint_state.position = [3.1415927410125732, 4.537856101989746, 5.93411922454834, -0.6108652353286743, 1.7453292608261108, -0.5235987901687622] start_state.joint_state = joint_state # goal_pose: # goal_orientation: # goal_pose = [self.object_z, -yi, xi - 0.05] # goal_orientation = [-0.00145713772037, -0.998970756926, 0.0364956710831, 0.0268955302573] z = self.object_z # goal_pose = [z + 0.02, -yi, xi] # Set the grasp pose: substract 17cm from the z value of the object centroid goal_pitches = [] goal_pitch = np.deg2rad(vfh[-1]) # goal_pitch = vfh[-1] + math.pi/2 goal_pitches.append(goal_pitch) # approaching_angle: vfh[-1] from the input for i in range(5): goal_pitches.append(goal_pitch + (i + 1) * (math.pi / 36)) goal_pitches.append(goal_pitch - (i + 1) * (math.pi / 36)) # Get the grasp orientation (currently the front direction) goal_orientations = [] for i in goal_pitches: goal_orientations.append(quaternion_from_euler(-i, math.radians(-5.0), math.radians(90.0), axes='rxyz')) # l = 0.001 l = 0.17 goal_poses = [] for i in goal_pitches: dx = math.sin(i - math.pi) * l dy = math.cos(i - math.pi) * l goal_poses.append([z + 0.2, -place_pose[1] + dx, place_pose[0] + dy]) feasibility1 = 0 i = 0 while not feasibility1 and i < len(goal_pitches): # CLF.add_box_client('can_bottom_x', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.1, 0.005, 0.005], 'red') # CLF.add_box_client('can_bottom_y', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.1, 0.005], 'blue') # CLF.add_box_client('can_bottom_z', [goal_poses[i][0] - 0.15, goal_poses[i][1], goal_poses[i][2]], goal_orientations[i], [0.005, 0.005, 0.1], 'green') [feasibility1, trajectory1] = CLF.move_goalpose_client('arm', 'gripper', start_state, goal_poses[i], goal_orientations[i], [], planner_name, n_attempt, c_time, n_repeat) i = i + 1 # time.sleep(1) # CLF.del_box_client('can_bottom_x') # CLF.del_box_client('can_bottom_y') # CLF.del_box_client('can_bottom_z') return [goal_poses[i], goal_orientations[i]] def back_home(self, current_joints): print "go to home" planner_name = 'RRTConnect' # planner_name = 'BiTRRT' n_attempt = 100 c_time = 3 n_repeat = 5 start_state = moveit_msgs.msg.RobotState() joint_state = sensor_msgs.msg.JointState() joint_state.header = std_msgs.msg.Header() joint_state.name = ['j2n6s300_joint_1', 'j2n6s300_joint_2', 'j2n6s300_joint_3', 'j2n6s300_joint_4', 'j2n6s300_joint_5', 'j2n6s300_joint_6'] joint_state.position = current_joints # joint_state.position = [3.1415927410125732, 4.537856101989746, 5.93411922454834, -0.6108652353286743, 1.7453292608261108, -0.5235987901687622] start_state.joint_state = joint_state # goal_pose: feasibility1 = 0 [feasibility1, trajectory1] = CLF.move_goalpose_client('arm', 'gripper', start_state, self.jaco_home_pos, self.jaco_home_ori, [], planner_name, n_attempt, c_time, n_repeat) def move_to(self, goal_pos): print "pos:", goal_pos[0], "ori:", goal_pos[1] CLF.move_cartesian('arm', goal_pos[0], goal_pos[1])