hybridrobotics / cbf-pointwise-feasibility

"Safety-Critical Control using Optimal-decay Control Barrier Functions with Guaranteed Point-wise Feasibility" by J. Zeng, B. Zhang, Z. Li and K. Sreenath https://arxiv.org/pdf/2103.12375.pdf

Feasibility with Control Barrier Functions

The infeasibility issues between safety constraints and input constraints might occur in safety-critical optimal control using control barrier functions, such CBF-QP and CLF-CBF-QP. We propose optimal-decay control barrier functions constraints that handles this infeasibility. This is the reference implementation of our paper:

Safety-Critical Control using Optimal-decay Control Barrier Functions with Guaranteed Point-wise Feasibility

PDF | Code | Figures

Jun Zeng, Bike Zhang, Zhongyu Li, and Koushil Sreenath

Citing

If you find this project useful in your work, please consider citing:

@inproceedings{zeng2021optimal-decay,
  title={Safety-Critical Control using Optimal-decay Control Barrier Functions with Guaranteed Point-wise Feasibility},
  author={Zeng, Jun and Zhang, Bike and Li, Zhongyu and Sreenath, Koushil},
  booktitle={2021 American Control Conference (ACC)},
  year={2021}
}

Instructions

We illustrate the problem about feasibility and safety using control barrier functions in this paper using an example of adpative cruise control.

• testFeasibility.m shows that infeasibility occurs in nominal CLF-CBF-QP and the proposed optimal-decay CLF-CBF-QP handles this issue.
• testHyperparameter.m shows that how hyperparameters influence the performance of safety.
• testSafety.m shows that control invariance might no longer be guaranteed if the initial condition is very challenging when the input constraints are introduced, which reveals the only point-wise feasibility (i.e. persistent feasible only for a subset of the safe set).
• testLieDerivative.m shows the calculation of lie derivatives for the example of adaptive cruise control.