def close(self, timestep): matplotrecorder.save_movie(f'img/av_swarm_{int(time.time())}.mp4', timestep)
# (5, 5, 1), # (3, 6, 2), # (3, 8, 2), # (3, 10, 2), # (7, 5, 2), # (9, 5, 2) # ] # [x,y,size(radius)] obstacleList = [(5, 5, 1), (4, 6, 1), (4, 8, 1), (4, 10, 1), (6, 5, 1), (7, 5, 1), (8, 6, 1), (8, 8, 1), (8, 10, 1)] # [x,y,size(radius)] # Set Initial parameters start = [0.0, 0.0, math.radians(0.0)] goal = [6.0, 7.0, math.radians(90.0)] rrt = RRT(start, goal, randArea=[-2.0, 15.0], obstacleList=obstacleList) path = rrt.Planning(animation=False) # Draw final path rrt.DrawGraph() plt.plot([x for (x, y) in path], [y for (x, y) in path], '-r') plt.grid(True) plt.pause(0.001) for i in range(10): matplotrecorder.save_frame() # save each frame plt.show() matplotrecorder.save_movie("animation.gif", 0.1)
def main(): parser = argparse.ArgumentParser() parser = argparse.ArgumentParser() parser.add_argument('--record', type=int, default=0) args = parser.parse_args() is_recorded = args.record train_dataset = SimpleDataset(total_num=250, is_confused=True, x=3, y=5, seed=1) test_dataset = SimpleDataset(total_num=100, is_confused=False, x=3, y=5, seed=2) model_PA = PassiveAggressive() model_PA_one = PassiveAggressiveOne(0.05) fig = plt.figure(figsize=(20, 4)) gs = gridspec.GridSpec(1, 10) fig_left = fig.add_subplot(gs[0, :3]) fig_right = fig.add_subplot(gs[0, 4:]) fig_left.set_xlim([ train_dataset.dataset.x1.min() - 0.1, train_dataset.dataset.x1.max() + 0.1 ]) fig_left.set_ylim([ train_dataset.dataset.x2.min() - 0.1, train_dataset.dataset.x2.max() + 0.1 ]) fig_left.set_title("Input Data & Trained Boundary", fontsize=15) fig_left.tick_params(labelsize=10) fig_right.set_xlim([0, len(train_dataset.y)]) fig_right.set_ylim([0, 1]) fig_right.set_title("Test Accuracy", fontsize=15) fig_right.tick_params(labelsize=10) fig_right.set_xlabel("Number of training data", fontsize=12) fig_right.set_ylabel("Accuracy", fontsize=12) line_x = np.array(range(-10, 10, 1)) line_y = line_x * 0 line_PA, = fig_left.plot(line_x, line_y, c="#2980b9", label="PA") line_PA_one, = fig_left.plot(line_x, line_y, c="#e74c3c", label="PA-1") fig_left.legend(handles=[line_PA, line_PA_one], fontsize=12) fig_right.legend(handles=[line_PA, line_PA_one], fontsize=12) valid_result_sample = [] accuracies_PA = [] accuracies_PA_one = [] imgs = [] for i in range(len(train_dataset.y)): fig_left.scatter(x=train_dataset.dataset.x1[i], y=train_dataset.dataset.x2[i], c=cm.cool(train_dataset.dataset.label[i]), alpha=0.5) model_PA.fit(train_dataset.feature_vec[i], train_dataset.y[i]) model_PA_one.fit(train_dataset.feature_vec[i], train_dataset.y[i]) accuracies_PA.append(test_dataset.valid_training_result(model_PA)) accuracies_PA_one.append( test_dataset.valid_training_result(model_PA_one)) a, b, c = model_PA.w line_y = (a * line_x + c) / (-b) line_PA.set_data(line_x, line_y) a, b, c = model_PA_one.w line_y = (a * line_x + c) / (-b) line_PA_one.set_data(line_x, line_y) fig_right.plot(accuracies_PA, c="#2980b9", label="PA") fig_right.plot(accuracies_PA_one, c="#e74c3c", label="PA-1") plt.pause(0.005) if is_recorded == 1: matplotrecorder.save_frame() if is_recorded == 1: matplotrecorder.save_movie("results.mp4", 0.005)