Repository for Robust Trajectory Optimization with Stochastic Complementarity
[Release Note Setpember 2020] This work has been submitted to RA-L and to arXiv
This repository contains code for robust contact-implicit trajectory optimization in the presence of uncertainty in the terrain model, as described in the paper Robust Trajectory Optimization over Uncertain Terrain with Stochastic Complementarity. Our implementation uses Expected Residual Minimization, a smoothing method for solving complementarity problems with uncertain data, and is tested in three examples including a footed hopping robot and a benchmarking experiment on sliding a block over terrain with uncertain friction characteristics.
- Authors: Luke Drnach and Ye Zhao
- Affiliation: The LIDAR Lab, Georgia Institute of Technology
This code was tested in MATLAB 2017a with Ubuntu 16.04 and a MATLAB version of Drake