This repository contains the python implementation for the iterative risk sensitive solvers accounting for process and measurement uncertainties published at the 2021 American Control Conference and freely available here
The scripts available under acc_2021/pneumatic_hopper
and acc_2021/point_cliff
should be sufficient to get started, here are few tips for the Pneumatic Hopper
- run
acc/pneumatic_hopper/solve_ddp.py
, this generates a plan and stores it inacc/pneumatic_hopper/solutions
. - similarly you can select the sensitivity parameter in
acc/pneumatic_hopper/config_pneumatic_hopper.py
and runacc/pneumatic_hopper/solve_firisk.py
to obtain and store the risk sensitive solutions. - once solutions are obtained, simulation with estimation and noise can be run with either
simulate_ilqg.py
orsimulate_risk_partial.py
If this code base is used in your research please cite the following paper
@inproceedings{hammoud2022irisc,
title={iRiSC: Iterative Risk Sensitive Control for Nonlinear Systems with Imperfect Observations},
booktitle={American Control Conference},
author={Hammoud, Bilal and Jordana, Armand and Righetti, Ludovic},
doi={10.23919/ACC53348.2022.9867200},
year={2022}
}