Multivehicle Flocking With Collision Avoidance via Distributed Model Predictive Control.

Area of Science:

• Robotics
• Control Systems Engineering
• Networked Systems

Background:

• Multivehicle systems have widespread applications, necessitating effective flocking control strategies.
• Existing methods often lack robust collision avoidance under limited communication ranges.
• Autonomous vehicle networks require advanced control for coordinated movement and safety.

Purpose of the Study:

• To develop a distributed flocking control strategy for autonomous vehicles with limited communication.
• To ensure collision avoidance as a primary condition for vehicles following a common trajectory.
• To provide sufficient conditions for the feasibility and stability of the proposed flocking control system.

Main Methods:

• Formulated the flocking control problem using centralized Model Predictive Control (MPC) with collision avoidance as an optimization constraint.
• Developed a Distributed Model Predictive Control (DMPC) approach based on controller consensus using the Alternating Direction Method of Multipliers (ADMM).
• Modified local controller constraints to guarantee collision avoidance within a finite number of ADMM iterations.

Main Results:

• The proposed DMPC strategy ensures that vehicles track a common desired trajectory without collisions.
• Feasibility and stability of the flocking control system were analyzed under practical conditions.
• Simulation and experimental results validated the effectiveness of the distributed flocking control method.

Conclusions:

• The developed distributed flocking control strategy effectively achieves stable trajectory tracking and collision avoidance for autonomous vehicles.
• The modified ADMM-based DMPC provides a feasible and stable solution for networked autonomous systems.
• This approach enhances the safety and coordination of multivehicle systems operating with limited communication.