Software

Robust Motion Planning & Non-linear MPC

This repository contains all the code required to implement the algorithms in the Robust motion planning paper from ICRA 2017.

There are currently four example systems within this repository: Planar VTOL (PVTOL), Flexible link robot (FLR), Quadrotor, and a two-state synthetic non-linear system (TubeMPC). Except for the quadrotor system which uses an implementation of the algorithm from Richter et. al., all other examples leverage the Tomlab Optimization Toolbox to solve for the optimal nominal motion plans/trajectories using the Pseudospectral method with Chebyshev-Gauss-Lobatto nodes. All examples also make use of the Tomlab toolbox to find the Riemann geodesics for computing the CCM-based feedback controller.

To solve for the control contraction metrics, we use the spotless Sum-of-squares optimization toolbox with the Mosek 8 solver. An additional function has also been provided to convert the solutions from the spotless toolbox into Matlab anonymous functions for direct implementation within the feedback controller.

Risk-sensitive Inverse Reinforcement Learning

This repository contains all the learning and evaluation code for the algorithms presented in the Risk-sensitive Inverse Reinforcement Learning via Semi- and Non-Parametric Methods paper. Additionally, we provide the raw and processed data files for all participants as well as the accompanying plotting functions to reproduce all tables and plots presented in the paper.

The learning and evaluation code is in MATLAB and the Vires VTD simulator code is in Python. The learning code dependencies are: MPT 3.0 Toolbox, Mosek 8 solver, and YALMIP parser.

Stabilizable Nonlinear Dynamics Learning (SNDL)

This repository contains all data generation, learning, and evaluation code for the Learning Stabilizable Dynamical Systems via Control Contraction Metrics paper. The repository also contains the data files containing the converged dynamics functions, test trajectories, and simulation results to replicate the plots presented in the paper.

All code is in MATLAB; please see the readme for all instructions.

Safe Planning with Model-Mismatch

This repository contains the SOS optimization and simulation code for the Robust Tracking with Model Mismatch for Fast and Safe Planning: an SOS Optimization Approach paper. Data files containing the converged V and K functions and simulation test environments for the results presented in the paper are provided for replicability.

All code is in MATLAB; please see the readme for all instructions.