openai / roboschool

DEPRECATED: Open-source software for robot simulation, integrated with OpenAI Gym.
Other
2.13k stars 488 forks source link

Apply pure force/torque and compute acceleration. #151

Closed charles-blouin closed 6 years ago

charles-blouin commented 6 years ago

I would like to use the roboschool framework for applied robotics. Specifically, for a quadcopter simulation. In practice, I will have access to the raw accelerometer and gyroscope data. Is there a way to obtain the accelerations without differentiating the position and orientation? Would additional python binding be needed?

Also, is there the equivalent of Mujoco qfrc_applied to apply a force and a torque on an object?

Thanks!

olegklimov commented 6 years ago

That seems to be a very simple environment. I'd recommend pybullet maybe for doing this, because you have immediate access to simulation internals from python.

charles-blouin commented 6 years ago

Thanks. After looking into it, seems to make more sense to use pybullet here. The pybullet function for applying a force is applyExternalForce. To obtain acceleration, it seems that the only way now is to differentiate the velocity.