Prediction and Control with Temporal Segment Models
This addresses the challenge of modeling dynamics in robotics or control systems, offering a novel approach for improved prediction and control, though it appears incremental as it builds on existing deep generative models.
The paper tackles the problem of learning dynamics for complex nonlinear systems by introducing a method based on deep generative models over temporal segments, which enables stable and accurate long-horizon predictions for stochastic systems like those with collisions, noise, and delays, and uses this for trajectory and policy optimization to evaluate performance and sample-efficiency.
We introduce a method for learning the dynamics of complex nonlinear systems based on deep generative models over temporal segments of states and actions. Unlike dynamics models that operate over individual discrete timesteps, we learn the distribution over future state trajectories conditioned on past state, past action, and planned future action trajectories, as well as a latent prior over action trajectories. Our approach is based on convolutional autoregressive models and variational autoencoders. It makes stable and accurate predictions over long horizons for complex, stochastic systems, effectively expressing uncertainty and modeling the effects of collisions, sensory noise, and action delays. The learned dynamics model and action prior can be used for end-to-end, fully differentiable trajectory optimization and model-based policy optimization, which we use to evaluate the performance and sample-efficiency of our method.