Resilient-Leader-Follower-CBF-QP-Controller

This repository contains the implementation of Strong $r$-robustness maintenance Control Barrier Function-based Quadratic Programming (CBF-QP) Controller. This guarantees the multi-robot network to maintain the sufficient number of connections among robots such a way that the robots can achieve leader-follower consensus despite the presence of misbehaving agents. Contrary to the previous works that assume robots can maintain predetermined graph topologies with known robustness properties, our CBF directly addresses robustness without imposing any fixed topology. This allows the robots to flexiblely form reconfigurable communication networks with desired levels of strong $r$-robustness while navigating in spatially restrictive environments. The algorithm also includes inter-agent and obstacle collision avoidances encoded as additional CBFs. We present three different scenarios, whose videos are shown below. For more information, please refer to our paper.

Hardware Experiment

Results for hardware experiments can be seen here.

Dependencies

This repository requires the following libraries: numpy, matplotlib, jax, and jaxlib.

For solving the CBF-QP, we need the following libraries: cvxpy and gurobi

Dynamics Model

We have adopted double integrator dynamics model to simulate robots' motions.

How to Run

In this repository, we have three different simulations: spread_out_sim, complex_sim1, and complex_sim2.

1. spread_out_sim shows the simulation of robots spreading out in an open space.
2. complex_sim1 and complex_sim2 show the simulation of robots going through some complex environments with obstacles.

To run each simulation scenario, you just need to run the corresponding python file.

spread_out_sim

complex_sim1

complex_sim2

For more Information...

Please check our Paper: https://arxiv.org/abs/2409.14675