dodox/fancy_gym
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge branch 'tt_cluster_debug' into 55-table-tennis-dev

Browse Source 
# Conflicts:
#	fancy_gym/black_box/black_box_wrapper.py
#	fancy_gym/envs/mujoco/box_pushing/box_pushing_env.py
#	fancy_gym/envs/mujoco/table_tennis/mp_wrapper.py
This commit is contained in:
Hongyi Zhou
2022-12-01 14:25:22 +01:00
parent a219f90d72 5750f6eb3d
commit a1d660d7ae
15 changed files with 476 additions and 350 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fancy_gym/examples/example_replanning_envs.py
+45 -21
View File
@@ -1,38 +1,62 @@
import fancy_gym
import numpy as np
import matplotlib.pyplot as plt
def plot_trajectory(traj):
plt.figure()
plt.plot(traj[:, 3])
plt.legend()
plt.show()
def run_replanning_envs(env_name="BoxPushingProDMP-v0", seed=1, iterations=1, render=True):
def example_run_replanning_env(env_name="BoxPushingDenseReplanProDMP-v0", seed=1, iterations=1, render=False):
env = fancy_gym.make(env_name, seed=seed)
env.reset()
for i in range(iterations):
done = False
desired_pos_traj = np.zeros((100, 7))
desired_vel_traj = np.zeros((100, 7))
real_pos_traj = np.zeros((100, 7))
real_vel_traj = np.zeros((100, 7))
t = 0
while done is False:
ac = env.action_space.sample()
obs, reward, done, info = env.step(ac)
desired_pos_traj[t: t + 25, :] = info['desired_pos']
desired_vel_traj[t: t + 25, :] = info['desired_vel']
# real_pos_traj.append(info['current_pos'])
# real_vel_traj.append(info['current_vel'])
t += 25
if render:
env.render(mode="human")
if done:
env.reset()
plot_trajectory(desired_pos_traj)
env.close()
del env
def example_custom_replanning_envs(seed=0, iteration=100, render=True):
# id for a step-based environment
base_env_id = "BoxPushingDense-v0"
wrappers = [fancy_gym.envs.mujoco.box_pushing.mp_wrapper.MPWrapper]
trajectory_generator_kwargs = {'trajectory_generator_type': 'prodmp',
'weight_scale': 1}
phase_generator_kwargs = {'phase_generator_type': 'exp'}
controller_kwargs = {'controller_type': 'velocity'}
basis_generator_kwargs = {'basis_generator_type': 'prodmp',
'num_basis': 5}
# max_planning_times: the maximum number of plans can be generated
# replanning_schedule: the trigger for replanning
# condition_on_desired: use desired state as the boundary condition for the next plan
black_box_kwargs = {'max_planning_times': 4,
'replanning_schedule': lambda pos, vel, obs, action, t: t % 25 == 0,
'condition_on_desired': True}
env = fancy_gym.make_bb(env_id=base_env_id, wrappers=wrappers, black_box_kwargs=black_box_kwargs,
traj_gen_kwargs=trajectory_generator_kwargs, controller_kwargs=controller_kwargs,
phase_kwargs=phase_generator_kwargs, basis_kwargs=basis_generator_kwargs,
seed=seed)
if render:
env.render(mode="human")
obs = env.reset()
for i in range(iteration):
ac = env.action_space.sample()
obs, reward, done, info = env.step(ac)
if done:
env.reset()
env.close()
del env
if __name__ == "__main__":
run_replanning_envs(env_name="BoxPushingDenseProDMP-v0", seed=1, iterations=1, render=False)
# run a registered replanning environment
example_run_replanning_env(env_name="BoxPushingDenseReplanProDMP-v0", seed=1, iterations=1, render=False)
# run a custom replanning environment
example_custom_replanning_envs(seed=0, iteration=8, render=True)
fancy_gym/examples/examples_movement_primitives.py
+12 -19
View File
@@ -17,8 +17,6 @@ def example_mp(env_name="HoleReacherProMP-v0", seed=1, iterations=1, render=True
# It takes care of seeding and enables the use of a variety of external environments using the gym interface.
env = fancy_gym.make(env_name, seed)
# env.traj_gen.basis_gn.show_basis(plot=True)
returns = 0
# env.render(mode=None)
obs = env.reset()
@@ -40,22 +38,16 @@ def example_mp(env_name="HoleReacherProMP-v0", seed=1, iterations=1, render=True
# Now the action space is not the raw action but the parametrization of the trajectory generator,
# such as a ProMP
ac = env.action_space.sample()
# ac[0] = 0.6866657733917236
# ac[1] = 0.08587364107370377
# This executes a full trajectory and gives back the context (obs) of the last step in the trajectory, or the
# full observation space of the last step, if replanning/sub-trajectory learning is used. The 'reward' is equal
# to the return of a trajectory. Default is the sum over the step-wise rewards.
print(f'target obs: {obs[-3:]}')
obs, reward, done, info = env.step(ac)
print(f'steps: {info["num_steps"][-1]}')
# Aggregated returns
returns += reward
if done:
# print(reward)
print(reward)
obs = env.reset()
print("=================New Episode======================")
# print("steps: {}".format(info["num_steps"][-1]))
def example_custom_mp(env_name="Reacher5dProMP-v0", seed=1, iterations=1, render=True):
@@ -163,21 +155,22 @@ def example_fully_custom_mp(seed=1, iterations=1, render=True):
if __name__ == '__main__':
render = True
render = False
# DMP
# example_mp("HoleReacherDMP-v0", seed=10, iterations=5, render=render)
example_mp("HoleReacherDMP-v0", seed=10, iterations=5, render=render)
# ProMP
# example_mp("HoleReacherProMP-v0", seed=10, iterations=5, render=render)
# example_mp("BoxPushingTemporalSparseProMP-v0", seed=10, iterations=1, render=render)
# example_mp("TableTennis4DProMP-v0", seed=10, iterations=10, render=True)
example_mp("HoleReacherProMP-v0", seed=10, iterations=5, render=render)
example_mp("BoxPushingTemporalSparseProMP-v0", seed=10, iterations=1, render=render)
example_mp("TableTennis4DProMP-v0", seed=10, iterations=20, render=render)
# ProDMP
# example_mp("BoxPushingDenseProDMP-v0", seed=10, iterations=16, render=render)
example_mp("TableTennis4DProDMP-v0", seed=10, iterations=5000, render=render)
# ProDMP with Replanning
example_mp("BoxPushingDenseReplanProDMP-v0", seed=10, iterations=4, render=render)
example_mp("TableTennis4DReplanProDMP-v0", seed=10, iterations=20, render=render)
example_mp("TableTennisWindReplanProDMP-v0", seed=10, iterations=20, render=render)
# Altered basis functions
# obs1 = example_custom_mp("Reacher5dProMP-v0", seed=10, iterations=1, render=render)
obs1 = example_custom_mp("Reacher5dProMP-v0", seed=10, iterations=1, render=render)
# Custom MP
# example_fully_custom_mp(seed=10, iterations=1, render=render)
example_fully_custom_mp(seed=10, iterations=1, render=render)