Torch implementation of MPPI Dependencies: similar as for the rnn repo Running: To run the model for one step, call the main file: python run_controller.py The controller works as follows: Instanciate a model, a cost, a controller and a observer. Call the controller on a fake state.