[Paper Review] Decentralized Goal Assignment and Trajectory Generation in Multi-Robot Networks.

Dimitra Panagou, Matthew Turpin+1 more|arXiv (Cornell University)|Feb 15, 2014

Distributed Control Multi-Agent Systems24 references3 citations

TL;DR

This paper proposes a decentralized method for goal assignment and trajectory generation in multi-robot networks using Lyapunov-like functions to select optimal, collision-free assignments locally. By combining switching strategies based on Lyapunov functions and barrier functions, the approach ensures safety, convergence, and scalability with minimal communication, validated through simulations in first-response scenarios.

ABSTRACT

Abstract — This paper considers the problem of decentralized goal assignment and trajectory generation for multi-robot networks when only local communication is available, and proposes an approach based on methods related to switched systems and set invariance. A family of Lyapunov-like functions is employed to encode the (local) decision making among candidate goal assignments, under which the agents pick the assignment which results in the shortest total distance to the goals. An additional family of Lyapunov-like barrier functions is activated in the case when the optimal assignment may lead to colliding trajectories, maintaining thus system safety while preserving the convergence guarantees. The proposed switching strategies give rise to feedback control policies which are scalable and computationally efficient as the number of agents increases, and therefore are suitable for applications including first-response deployment of robotic networks under limited information sharing. Simulations demonstrate the efficacy of the proposed method. I.

Motivation & Objective

• To address decentralized goal assignment and trajectory generation in multi-robot networks under limited local communication.
• To ensure system safety by preventing collisions during trajectory execution despite optimal assignment choices.
• To maintain convergence guarantees and scalability as the number of agents increases.
• To develop computationally efficient feedback control policies suitable for real-time deployment.
• To enable first-response robotic network deployments with minimal information sharing.

Proposed method

• The method employs a family of Lyapunov-like functions to encode local decision-making for selecting goal assignments that minimize total distance to goals.
• Switching strategies are derived from these Lyapunov functions to dynamically update assignments based on local information.
• An additional family of Lyapunov-like barrier functions is activated when optimal assignments risk trajectory collisions, ensuring safety.
• The switching logic integrates both function families to preserve convergence while enforcing collision avoidance.
• Feedback control policies are synthesized from the switching strategy, enabling scalable and distributed execution.
• The approach is designed to be computationally efficient and suitable for large-scale multi-robot systems.

Experimental results

Research questions

• RQ1 How can decentralized goal assignment be achieved in multi-robot networks with only local communication?
• RQ2 What control strategy ensures convergence to optimal assignments while avoiding collisions?
• RQ3 How can safety be maintained without centralized coordination or global information?
• RQ4 What is the scalability of the proposed method as the number of agents increases?
• RQ5 Can the method be applied effectively in first-response robotic network deployments?

Key findings

• The proposed method achieves decentralized goal assignment that minimizes total distance to goals using only local information and communication.
• The integration of barrier functions ensures collision-free trajectories even when optimal assignments would otherwise lead to conflicts.
• The switching strategy based on Lyapunov functions guarantees convergence to a stable assignment under the proposed control policy.
• The method scales efficiently with the number of agents due to its local computation and communication requirements.
• Simulations confirm the efficacy of the approach in first-response deployment scenarios with limited information sharing.
• The feedback control policies are computationally efficient and suitable for real-time implementation in large robotic networks.