Playing with MuJoCo is a lot of fun. I recommend it for people who just want to experiment with RL. You can do random weird stuff like this: https://www.youtube.com/watch?v=e-9u_vepfY8. I'm really glad DeepMind is continuing to update it.
Holy Molly. The live demo where you put a place holder and the trained robot dog just goes there is insane! You can also apply force (aka kick) the robot while it is moving!
Really cool. If we could do sim training to rl conversion in robotics, we migth get a llm moment for robotics, with good pretrained movement algorithms.
Playing with MuJoCo is a lot of fun. I recommend it for people who just want to experiment with RL. You can do random weird stuff like this: https://www.youtube.com/watch?v=e-9u_vepfY8. I'm really glad DeepMind is continuing to update it.
Holy Molly. The live demo where you put a place holder and the trained robot dog just goes there is insane! You can also apply force (aka kick) the robot while it is moving!
An open-source library for GPU-accelerated robot learning and sim-to-real transfer.
Really cool. If we could do sim training to rl conversion in robotics, we migth get a llm moment for robotics, with good pretrained movement algorithms.
This is great - I haven't looked at the documentation yet but do you support soft body simulation?
MuJoCo does support soft bodies:
https://mujoco.readthedocs.io/en/latest/modeling.html#deform...
The GPU part of MuJoCo only supports soft collisions rather than fully deformable bodies.