Robot walking backwards in headless mode

[Dzhange]

Hi

After I fixed the headless rendering issue I had another problem.

While running the policy_player in headless mode, I turned render option to False when initializing the environment, and used a gym.wrappers.monitoring.video_recorder to record the scene.
Here you can see the robot is walking smoothly backwards(walking environment, in mounts terrain).

2021-05-19.10.56.13.996819.mp4

Another case(gallop, plane), but this time is turning around

2021-05-19.11.09.19.051276.mp4

I tried enabling the sleep control in

rex-gym/rex_gym/envs/rex_gym_env.py

Lines 387 to 398 in 2666304

	if self._is_render:
	# Sleep, otherwise the computation takes less time than real time,
	# which will make the visualization like a fast-forward video.
	time_spent = time.time() - self._last_frame_time
	self._last_frame_time = time.time()
	time_to_sleep = self.control_time_step - time_spent
	if time_to_sleep > 0:
	time.sleep(time_to_sleep)
	base_pos = self.rex.GetBasePosition()
	# Keep the previous orientation of the camera set by the user.
	[yaw, pitch, dist] = self._pybullet_client.getDebugVisualizerCamera()[8:11]
	self._pybullet_client.resetDebugVisualizerCamera(dist, yaw, pitch, base_pos)

but the problem still exists. Any ideas?

Thanks!

[nicrusso7]

this looks like a timing issue, and I see you had my same impression (as you tried to edit the simulation timestep timing. However that block won't be executed if render = False, did you tried removing that if statement as well?

[Dzhange]

Yeah I tried to remove that if statement, actually the videos are recorded with this block executed.

I'm wondering if this problem is related to the parameters updated here?

rex-gym/rex_gym/envs/gym/walk_env.py

Lines 262 to 275 in 2666304

	if self._is_render and self._is_debug:
	position, orientation, step_length, step_rotation, step_angle, step_period = \
	self._read_inputs(base_pos_coeff, gait_stage_coeff)
	else:
	position = np.array([base_x,
	self._base_y * base_pos_coeff,
	self._base_z * base_pos_coeff])
	orientation = np.array([self._base_roll * base_pos_coeff,
	self._base_pitch * base_pos_coeff,
	self._base_yaw * base_pos_coeff])
	step_length = (self.step_length if self.step_length is not None else step) * gait_stage_coeff
	step_rotation = (self.step_rotation if self.step_rotation is not None else 0.0)
	step_angle = self.step_angle if self.step_angle is not None else 0.0
	step_period = (self.step_period if self.step_period is not None else period)

[nicrusso7]

I may be wrong but, nope, don't think so.. in both cases, the values we set are actually the same

rex-gym/rex_gym/envs/gym/walk_env.py

Line 156 in 2666304

def setup_ui(self, base_x, step, period):

rex-gym/rex_gym/envs/rex_gym_env.py

Line 260 in 2666304

self._ranges = {

[nicrusso7]

Do you have the same behavior running the policy_player in render mode? just wondering if the pybullet update made some changes

[Dzhange]

HIii, just checked several things:

1. I found that walking backwards or forwards is randomly chosen, so I think the main problem is the gait/frequency etc
2. sometimes the step_length parameter varies a bit(i.e. from -0.1 to -0.3), not sure if this is normal?
3. I tried the newest pybullet with a monitor, and the robot walked well. I also tested the session with 3 settings:

1. render and record video at same time --> moving slowly

out.mp4

2. render only --> normal

out1.mp4

3)record video only --> rapid frequency, yet slow

cap_only.mp4

I guess the problem is mainly about timing, let me take a deeper look at that.
Here is my implementation of recording if you are interested:
https://github.com/Dzhange/rex-gym/blob/779470994e74259f5b29db62beb4f4aa05f8ddeb/rex_gym/playground/policy_player.py#L45-L84

[Dzhange]

I think I found the problem:

In render mode, fps is around 50
but when using capture_frame to visualize, the fps drops to ~3

I'm wondering if there is any method to fix the time for each step in the pybullet world?
or do you have any suggestions?
Thanks!

[nicrusso7]

Each env has its own metadata:

rex-gym/rex_gym/envs/gym/walk_env.py

Line 27 in 2666304

metadata = {"render.modes": ["human", "rgb_array"], "video.frames_per_second": 66}

Have you tried editing video.frames_per_second?

[Dzhange]

Hi,

thanks for your advice, this did make the video more nature!
After some struglling I raised the fps to ~20 by adding this block

https://github.com/Dzhange/rex-gym/blob/3ffb674ade13ceac3352c3329f1804eb21a08403/rex_gym/envs/rex_gym_env.py#L234-L242

not sure if this would support batchified training but I think that has little to do with this issue, I'll close itHi,

thanks for your advice, this did make the video more nature!
After some struglling I raised the fps to ~20 by adding this block

            self._pybullet_client = bullet_client.BulletClient(connection_mode=pybullet.DIRECT)
            import pkgutil
            egl = pkgutil.get_loader('eglRenderer')
            if egl:
                pluginId = pybullet.loadPlugin(egl.get_filename(), "_eglRendererPlugin")
            else:
                pluginId = pybullet.loadPlugin("eglRendererPlugin")
            print("pluginId=",pluginId)

not sure if this fps would support batchified training, but I think that has little to do with this issue, I'll close it