OpenRobotLab / HIMLoco

Learning-based locomotion control from OpenRobotLab, including Hybrid Internal Model & H-Infinity Locomotion Control
https://junfeng-long.github.io/HIMLoco/
Other
269 stars 29 forks source link

Are you planning to release code for Orbit #9

Open dhananjayrajmane opened 5 months ago

dhananjayrajmane commented 5 months ago

Hi

are you planning to release the code for orbit. I am trying to replicate your work in orbit but facing few issues in development. If possible can you help me there ?

Junfeng-Long commented 5 months ago

Hi, thank you for your interest in our work. Actually, we haven't started using Orbit. But we would like to help in our free time if you can provide more information.

dhananjayrajmane commented 5 months ago

@Junfeng-Long Thank for quick reply. I am able to collect both observations and critic/priv observations through ObservationManager but i am facing issue while implementing reward_foot_clearance.

Similarly for step() function : Orbit return observation, priv_observation, rewards , done and extras. Required guidance for getting termination_ids and termination observations.

dhananjayrajmane commented 2 months ago

@Junfeng-Long I am able to replicate and implement your paper in Isaac lab but while training mean reward is negative till iterations 1000 but in isaacgym training mean reward came positive too early. Can you help me to replicate it