Distributed Estimation and Task-Driven Relational Formation Reconfiguration in Soccer Teams

This technical report was developed as a final course project for AAE59000MAAC: Multi-Agent Autonomy and Control at Purdue University. The report studies soccer formation coordination from a multi-agent control perspective. Instead of treating formation as a fixed absolute template, it models a soccer team formation as a task-dependent relational structure represented by tactical graph edges, desired distances, and role-dependent constraints.

The report proposes a distributed estimation-control framework in which each player estimates a latent team state from local observations, updates a task-dependent tactical graph, and performs projected local control to reduce formation error while respecting role feasibility. A small five-player defensive-block simulation is included to illustrate defensive compactness recovery, fullback-pressing reconfiguration, and distributed estimation agreement.

This record includes the final technical report and the MATLAB script used to generate the simulation results.