def main():
idx = input('Please choose the controller that you want to sample -- 1. DS, 2. IL, 3. RRT: ')
try:
idx = int(idx)
assert 1 <= idx <= 3
except:
print('Invalid input! The input needs to be a number between 1 and 3. ')
quit()
if idx == 1:
env = RBF2dGymEnv(time_limit=500, oob_termination=False, use_lidar=False)
controller = DSController()
controller_kernel = TransitionKernel()
elif idx == 2:
env = RBF2dGymEnv(use_lidar=True)
controller = ILController()
controller_kernel = TransitionKernel()
else:
env = RBF2dGymEnv(use_lidar=False)
controller = RRTController()
controller_kernel = RRTKernelNormal([-1, -1], [1, 1])
while True:
env.reset(np.random.uniform(low=-0.7, high=0.7, size=(15, 2)))
traj = controller.get_trajectory(env, controller_kernel)
plt.imshow(env.arena.occ_grid, origin='lower', extent=[-1.2, 1.2, -1.2, 1.2],
cmap='gray_r', vmin=0, vmax=2)
plt.plot(traj[:, 0], traj[:, 1])
plt.title('Close this window to view next obstacle configuration. ')
plt.show()
def main():
idx = input(
'Please select the controller -- \n1a. Original Dynamical System\n1b. Improved Dynamical System\n'
+
'2. Reinforcement Learning\n3. Rapidly-Exploring Random Tree\nEnter your choice: '
)
assert idx in ['1a', '1b', '2', '3'
], 'Invalid input! The input needs to be 1a, 1b, 2, or 3. '
env = PandaEnv()
env.render()
env_kernel = ReachingEnvKernelNormal()
if idx == '1a':
controller = DSController(typ='original', visualize=True)
controller_kernel = TransitionKernel()
elif idx == '1b':
controller = DSController(typ='improved', visualize=True)
controller_kernel = TransitionKernel()
elif idx == '2':
controller = RLController(visualize=True)
controller_kernel = TransitionKernel()
else:
controller = RRTController(visualize=True)
controller_kernel = RRTKernelNormal()
while True:
env_kernel.sample_prior()
controller_kernel.sample_prior()
env.reset(target_loc=env_kernel.value)
controller.get_trajectory(env, controller_kernel)
def main():
idx = input('Please select the controller -- \n1. Dynamical System\n2. Imitation Learning\n3. Rapidly-Exploring Random Tree\n' +
'Enter a number between 1 and 3: ')
try:
idx = int(idx)
assert 1 <= idx <= 3
except:
print('Invalid input! The input needs to be a number between 1 and 3. ')
quit()
if idx == 1:
env = RBF2dGymEnv(time_limit=500, oob_termination=False, use_lidar=False)
controller = DSController()
controller_kernel = TransitionKernel()
save_fn = 'samples/ds_prior.pkl'
elif idx == 2:
env = RBF2dGymEnv(use_lidar=True)
controller = ILController()
controller_kernel = TransitionKernel()
save_fn = 'samples/il_prior.pkl'
else:
env = RBF2dGymEnv(use_lidar=False)
controller = RRTController()
controller_kernel = RRTKernelNormal([-1, -1], [1, 1])
save_fn = 'samples/rrt_prior.pkl'
env_kernel = RBF2dEnvKernelNormal()
sample_prior(2000, save_fn, env, controller, env_kernel, controller_kernel)
def main():
idx = input(
'Please choose the controller that you want to sample -- 1. DS, 2. IL, 3. RRT: '
)
try:
idx = int(idx)
assert 1 <= idx <= 3
except:
print(
'Invalid input! The input needs to be a number between 1 and 3. ')
quit()
if idx == 1:
env = RBF2dGymEnv(time_limit=500,
oob_termination=False,
use_lidar=False)
controller = DSController()
controller_kernel = TransitionKernel()
save_fn = 'samples/ds_prior.pkl'
elif idx == 2:
env = RBF2dGymEnv(use_lidar=True)
controller = ILController()
controller_kernel = TransitionKernel()
save_fn = 'samples/il_prior.pkl'
else:
env = RBF2dGymEnv(use_lidar=False)
controller = RRTController()
controller_kernel = RRTKernelNormal([-1, -1], [1, 1])
save_fn = 'samples/rrt_prior.pkl'
env_kernel = RBF2dEnvKernelNormal()
sample_prior(2000, save_fn, env, controller, env_kernel, controller_kernel)
def main():
idx = input(
'Please select the controller type: \n1. Dynamical System\n2. Imitation Learning\n3. Rapidly-Exploring Random Tree\n'
+ 'Enter a number between 1 and 3: ')
try:
idx = int(idx)
assert 1 <= idx <= 3
except:
print(
'Invalid input! The input needs to be a number between 1 and 3. ')
quit()
if idx == 1:
env = RBF2dGymEnv(time_limit=500,
oob_termination=False,
use_lidar=False)
controller = DSController()
controller_kernel = TransitionKernel()
elif idx == 2:
env = RBF2dGymEnv(use_lidar=True)
controller = ILController()
controller_kernel = TransitionKernel()
else:
env = RBF2dGymEnv(use_lidar=False)
controller = RRTController()
controller_kernel = RRTKernelNormal([-1, -1], [1, 1])
env_kernel = RBF2dEnvKernelNormal()
while True:
env_kernel.sample_prior()
controller_kernel.sample_prior()
env.reset(env_kernel.value)
traj = controller.get_trajectory(env, controller_kernel)
plt.imshow(env.arena.occ_grid,
origin='lower',
extent=[-1.2, 1.2, -1.2, 1.2],
cmap='gray_r',
vmin=0,
vmax=2)
plt.plot(traj[:, 0], traj[:, 1])
plt.title('Close this window to view next obstacle configuration. ')
plt.show()
def ds_improved_min_end_distance():
name = f'samples/{inspect.currentframe().f_code.co_name}.pkl'
samples = sample(N=10000,
alpha=0.1,
prior_file='samples/ds_improved_prior.pkl',
N_sigma=1000,
behavior_func=behavior.ee_distance_behavior,
env=PandaEnv(),
env_kernel=ReachingEnvKernelNormal(),
controller=DSController(typ='improved'),
controller_kernel=TransitionKernel(),
target_type='match',
target_behavior=0,
save=name)
def main():
idx = input('Please select the controller -- \n1a. Original Dynamical System\n1b. Improved Dynamical System\n' +
'2. Reinforcement Learning\n3. Rapidly-Exploring Random Tree\nEnter your choice: ')
assert idx in ['1a', '1b', '2', '3'], 'Invalid input! The input needs to be 1a, 1b, 2, or 3. '
env = PandaEnv()
env_kernel = ReachingEnvKernelNormal()
if idx == '1a':
controller = DSController(typ='original')
controller_kernel = TransitionKernel()
save_fn = 'samples/ds_original_prior.pkl'
elif idx == '1b':
controller = DSController(typ='improved')
controller_kernel = TransitionKernel()
save_fn = 'samples/ds_improved_prior.pkl'
elif idx == '2':
controller = RLController()
controller_kernel = TransitionKernel()
save_fn = 'samples/rl_prior.pkl'
else:
controller = RRTController()
controller_kernel = RRTKernelNormal()
save_fn = 'samples/rrt_prior.pkl'
sample_prior(2000, save_fn, env, controller, env_kernel, controller_kernel)