Real-Time Physics Simulation Forum 2022-07-29T16:44:07+00:00 https://pybullet.org/Bullet/phpBB3/app.php/feed/topic/13463 2022-07-29T16:44:07+00:00 2022-07-29T16:44:07+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=44070#p44070 <![CDATA[Re: InvertedPendulumBulletEnv-v0 observation dimension meaning?]]> https://github.com/bulletphysics/bullet ... ula.py#L48.

np.array([x, vx, np.cos(self.theta), np.sin(self.theta), theta_dot])

It's the same four-dimensional space, except that rather than cart position, cart velocity, pole angle, pole angular velocity like usual, it's the cos and sin of pole angle.

Statistics: Posted by mretchin — Fri Jul 29, 2022 4:44 pm

]]> 2022-07-29T01:54:33+00:00 2022-07-29T01:54:33+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=44069#p44069 <![CDATA[InvertedPendulumBulletEnv-v0 observation dimension meaning?]]> https://github.com/bulletphysics/bullet ... um_envs.py) is 5-dimensional, which is different from the usual 4-dimensional observation vector: [cart position, cart velocity, pendulum angle, pendulum angular velocity]. What do the observation dimensions correspond to?

I'd like to start my episode with the pendulum facing down, but it's difficult to do that if I don't know what the observation space is.

Thanks!

Statistics: Posted by mretchin — Fri Jul 29, 2022 1:54 am

]]>