GrpAvoid: Multigroup Collision-Avoidance Control and Optimization for UAV Swarm.

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Traditional collision-avoidance for Unmanned Aerial Vehicle (UAV) swarms focus on individual units, lacking strategies for multi-group interactions.
• Existing models can lead to deadlock situations in complex scenarios, hindering UAV swarm progress.

Purpose of the Study:

• To develop a novel modeling and optimization approach for collision avoidance in multi-group UAV scenarios.
• To address and resolve deadlock situations inherent in current multi-group UAV coordination methods.

Main Methods:

• Introduced group-level collision detection and an adaptation mechanism for UAV groups.
• Developed a two-level control model for inter-group and intra-group collision avoidance.
• Utilized evolutionary multitask optimization and adaptive fitness evaluation for parameter calibration.

Main Results:

• Demonstrated superior performance in deadlock resolution, motion stability, and distance maintenance in multigroup UAV scenarios.
• Achieved significant reduction in execution time for computationally intensive optimization processes.
• Validated the model's effectiveness compared to state-of-the-art collision-avoidance techniques.

Conclusions:

• The proposed approach effectively resolves deadlocks and enhances collision avoidance in complex, multi-group UAV environments.
• The optimization method significantly reduces computational overhead, making it practical for real-world applications.
• This work provides a robust framework for safe and efficient large-scale UAV swarm operations.