A Caltech Library Service

Chance-Constrained Trajectory Optimization for Safe Exploration and Learning of Nonlinear Systems

Nakka, Yashwanth Kumar and Liu, Anqi and Shi, Guanya and Anandkumar, Anima and Yue, Yisong and Chung, Soon-Jo (2020) Chance-Constrained Trajectory Optimization for Safe Exploration and Learning of Nonlinear Systems. . (Unpublished)

[img] PDF (27 Oct 2020) - Submitted Version
See Usage Policy.


Use this Persistent URL to link to this item:


Learning-based control algorithms require collection of abundant supervision for training. Safe exploration algorithms enable this data collection to proceed safely even when only partial knowledge is available. In this paper, we present a new episodic framework to design a sub-optimal pool of motion plans that aid exploration for learning unknown residual dynamics under safety constraints. We derive an iterative convex optimization algorithm that solves an information-cost Stochastic Nonlinear Optimal Control problem (Info-SNOC), subject to chance constraints and approximated dynamics to compute a safe trajectory. The optimization objective encodes both performance and exploration, and the safety is incorporated as distributionally robust chance constraints. The dynamics are predicted from a robust learning model. We prove the safety of rollouts from our exploration method and reduction in uncertainty over epochs ensuring consistency of our learning method. We validate the effectiveness of Info-SNOC by designing and implementing a pool of safe trajectories for a planar robot.

Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription Paper
Nakka, Yashwanth Kumar0000-0001-7897-3644
Yue, Yisong0000-0001-9127-1989
Chung, Soon-Jo0000-0002-6657-3907
Additional Information:This work was in part funded by the Jet Propulsion Laboratory, California Institute of Technology and the Raytheon Company. We would like to thank Irene S. Crowell for her contribution to the implementation of the projection method.
Group:GALCIT, Center for Autonomous Systems and Technologies (CAST)
Funding AgencyGrant Number
Raytheon CompanyUNSPECIFIED
Record Number:CaltechAUTHORS:20200526-150616242
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:103472
Deposited By: Tony Diaz
Deposited On:26 May 2020 22:31
Last Modified:06 Nov 2020 21:38

Repository Staff Only: item control page