def get_optimal(theta):
opt = TrajOpt(theta=theta)
xi, res, solve_time = opt.optimize()
T = 10.0
traj = Trajectory(xi, T)
env = Task()
state = env.reset()
start_time = time.time()
curr_time = time.time() - start_time
while curr_time < T + 0.5:
q_des = traj.get(curr_time)
qdot = 10 * (q_des - state["joint_position"][0:7])
next_state, reward, done, info = env.step(qdot)
state = next_state
curr_time = time.time() - start_time
env.close()
return xi
def replay_demo(xi):
T = 10.0
traj = Trajectory(xi, T)
env = Task()
state = env.reset()
count = 0
max_count = 10001
timesteps = np.linspace(0, T, max_count)
xi = []
while count < max_count:
curr_time = timesteps[count]
if count % 1000 == 0:
xi.append(state["joint_position"][0:7].tolist())
count += 1
q_des = traj.get(curr_time)
qdot = 10 * (q_des - state["joint_position"][0:7])
next_state, reward, done, info = env.step(qdot)
state = next_state
env.close()
return xi
def main():
theta = float(sys.argv[1])
opt = TrajOpt(theta=theta)
xi, res, solve_time = opt.optimize()
opt.trajcost(xi)
print("it took me this long to solve: ", solve_time)
T = 10.0
traj = Trajectory(xi, T)
env = Task()
state = env.reset()
start_time = time.time()
curr_time = time.time() - start_time
while curr_time < T + 0.5:
q_des = traj.get(curr_time)
qdot = 10 * (q_des - state["joint_position"][0:7])
next_state, reward, done, info = env.step(qdot)
state = next_state
curr_time = time.time() - start_time
env.close()