This repository contains the code for the paper "Human–Machine Shared Stabilization Control Based on Safe Adaptive Dynamic Programming With Bounded Rationality" published in the International Journal of Robust and Nonlinear Control Journal (March 2025). The simulation demonstrates the basic idea of human-machine shared control based on safe ADP with Level-$k$ bounded rationality.
- Details of the paper can be found at: Junkai Tan's Publications
- Download the paper at: Download pdf
If you use this code or find our research helpful, please cite our paper:
@article{Tan2025,
author = {Tan, Junkai and Wang, Jie and Xue, Shuangsi and Cao, Hui and Li, Haojun and Guo, Zhongxin},
title = {Human–Machine Shared Stabilization Control Based on Safe Adaptive Dynamic Programming With Bounded Rationality},
journal = {International Journal of Robust and Nonlinear Control},
year = {2025},
month = {March},
doi = {10.1002/rnc.7931},
url = {https://doi.org/10.1002/rnc.7931}
}This article addresses shared control between bounded rational humans and cooperative autonomous machines. The key focus is ensuring safety in human-machine interactions through:
- A barrier-function-based state transformation that enforces full state safety constraints
- A level-k thinking framework to model bounded rationality
- Adaptive dynamic programming to approximate each level-k control policy
- A Softmax-based probabilistic distribution to model human behavior and its inherent uncertainty
The shared control framework combines human and machine inputs to achieve safe, efficient stabilization. Simulation results demonstrate that:
- Full-state asymmetric constraints and stabilization are maintained in safety-critical situations
- System safety is preserved even when one participant lacks safety awareness
Files:
Simulation/main.m: Main file to run the simulationSimulation/plot_fig.m: Main file to plot the resultsSimulation/plot_bar.m: Plot the distribution of rationality level