Bug: function (update_root_state_tensor_by_obstacles_tensor) does not work properly

tud-airlab / mppi-isaac

Model Predictive Path Integral Control using isaacgym for rollouts, gpu-accelerated

178 stars 21 forks source link

Bug: function (update_root_state_tensor_by_obstacles_tensor) does not work properly #50

Open yj-Tang opened 11 months ago

yj-Tang commented 11 months ago

I tested the update_root_state_tensor_by_obstacles_tensor function in the isaacgym wrapper. I think there is something wrong with it. Tested it here https://github.com/yj-Tang/mppi-isaac/blob/82420f22dbe64a7aa6cb9d7256c78efadb3132b5/examples/boxer_push_server.py#L173C1-L207C89. When I set the pos of the obstacles, they did not change as I expected

cpezzato commented 11 months ago

I think the problem is in isaacgym_wrapper.py. the indexes of the obstacles do not seem to be set correctly in update_root-state_tensor_by_obstacle_tensor.

I will have a look