Risk-Averse Zero-Order Trajectory OptimizationDownload PDF

19 Jun 2021, 10:04 (modified: 05 Nov 2021, 07:26)CoRL2021 PosterReaders: Everyone
Keywords: CEM, data-driven MPC, uncertainty, model-based RL
Abstract: We introduce a simple but effective method for managing risk in zero-order trajectory optimization that involves probabilistic safety constraints and balancing of optimism in the face of epistemic uncertainty and pessimism in the face of aleatoric uncertainty of an ensemble of stochastic neural networks. Various experiments indicate that the separation of uncertainties is essential to performing well with data-driven MPC approaches in uncertain and safety-critical control environments.
Supplementary Material: zip
Poster: pdf
11 Replies