def plot_avg_planning_time_vs_trust_region(self, results):
# results = list(self.db.find({"type": "kuka_only_random_trust_region"}))
# result = list(self.db.find({"type": "donbot_arm_only_random_trust_region"}))
# print len(result)
trust_region = DefaultOrderedDict(list)
plan_t = DefaultOrderedDict(list)
sol_t = DefaultOrderedDict(list)
col_t = DefaultOrderedDict(list)
prb_t = DefaultOrderedDict(list)
for res in results:
trust = res["solver_config"]["trust_region_size"]
trust_region[trust].append(trust)
plan_t[trust].append(res["planning_time"])
sol_t[trust].append(res["solving_time"])
col_t[trust].append(res["collision_check_time"])
prb_t[trust].append(res["prob_model_time"])
avg_planning_time = []
avg_solving_time = []
avg_collision_check_time = []
avg_prob_model_time = []
trust_region = OrderedDict(sorted(trust_region.items()))
for k in trust_region:
avg_planning_time.append(sum(plan_t[k]) / len(plan_t[k]))
avg_solving_time.append(sum(sol_t[k]) / len(sol_t[k]))
avg_collision_check_time.append(sum(col_t[k]) / len(col_t[k]))
avg_prob_model_time.append(sum(prb_t[k]) / len(prb_t[k]))
ys = [avg_planning_time, avg_solving_time, avg_collision_check_time, avg_prob_model_time]
xs = [trust_region.keys()] * len(ys)
labels = ["Planning time", "Solving time", "Collision check time", "Problem modelling time"]
plotter.multi_plot_best_fit_curve(xs, ys, labels, "Average time taken vs Trust region size", "Trust region size", "Average Time (s)",
deg=2)
def iterations_vs_trust_region(self, results):
trust_region = DefaultOrderedDict(list)
qp_iters = DefaultOrderedDict(list)
sqp_iters = DefaultOrderedDict(list)
for res in results:
trust = res["solver_config"]["trust_region_size"]
trust_region[trust].append(trust)
qp_iters[trust].append(res["num_qp_iterations"])
sqp_iters[trust].append(res["num_sqp_iterations"])
avg_qp_iters = []
avg_sqp_iters = []
trust_region = OrderedDict(sorted(trust_region.items()))
for k in trust_region:
avg_qp_iters.append(sum(qp_iters[k]) / len(qp_iters[k]))
avg_sqp_iters.append(sum(sqp_iters[k]) / len(sqp_iters[k]))
ys = [avg_qp_iters, avg_sqp_iters]
xs = [trust_region.keys()] * len(ys)
labels = ["QP iterations", "SQP iterations"]
plotter.multi_plot_best_fit_curve(xs, ys, labels, "Number of SQP and QP iterations vs Trust region", "Trust region", "Number of iterations",
deg=2)