Model-Based Imitation Learning

Trajectory Optimization

Class of planning algorithms that iteratively improve an entire trajectory using gradients from the reward and dynamic models, as opposed to value-based methods.

← 뒤로