Swarm Coordination Under Conflict

We propose different models for interactions between two antagonistic multi-agent swarms, pursuers and evaders. Different control techniques are deployed in the pursuers to help them capturing the evaders.

In this page we present some animations corresponding to the example cases presented in the publications linked in each case.

First Order Model

Undamped example case: there is no extra control term applied to the pursuers (black squares) to help them in the task of capturing the evaders (red circles). The dynamics of the operation guides the interaction.

Damped example case: an extra control term is applied to the pursuers to expedite the captur of the evaders (red circles). Chatter is noticeable in the pursuers.

The lyapunov-based control term is sustained and gradually reduced in order to alleviate the chatter.

Second order Model

The agents are now modeled by second order dynamics. A PD Control guarantees the total error between pursuers and evaders is bounded. Due to uncertainties in the dynamics, the error bound is large.

A sliding mode controller is deployed into pursuers. The forces in the evaders and some parametric variations are considereded as uncertainties. The boundary layer concept is deployed to alleviate the control chatter.