So, to achieve multithread dynamics rollout (as MPPI implemented), one important thing to test out is integrating CUDA-supported API with MPC dynamics rollout from f1tenth lab8. To start, we will use the f1tenth lab8 repo as a starting point and swap out the dynamics part with CuPy data structures to see if it works! Sounds easy!
So, to achieve multithread dynamics rollout (as MPPI implemented), one important thing to test out is integrating CUDA-supported API with MPC dynamics rollout from f1tenth lab8. To start, we will use the f1tenth lab8 repo as a starting point and swap out the dynamics part with CuPy data structures to see if it works! Sounds easy!