Rolling horizon coverage control with collaborative autonomous agents.

Area of Science:

• Robotics and Control Systems
• Computer Vision
• Optimization Theory

Background:

• Collaborative coverage planning is crucial for efficient multi-agent tasks.
• Non-myopic planning and visibility assessment are key challenges in aerial surveillance.
• Reducing redundancy in distributed autonomous systems is an ongoing research area.

Purpose of the Study:

• To develop a distributed model predictive control framework for aerial agent coverage.
• To enable collaborative, non-myopic coverage planning over a finite horizon.
• To integrate complex visibility constraints into an optimization framework for 3D object inspection.

Main Methods:

• Formulation of the collaborative coverage problem as a distributed model predictive control problem.
• Integration of inter-agent constraints to minimize work redundancy.
• Utilizing light-path propagation and logical expressions for non-linear visibility assessment.
• Embedding binary constraints within a mixed-integer optimization framework.

Main Results:

• Demonstrated a novel coverage controller for distributed autonomous aerial agents.
• Successfully optimized agents' motion and camera control for collaborative coverage.
• Validated the approach through simulations and practical building inspection using unmanned aerial vehicles (UAVs).

Conclusions:

• The proposed controller enables efficient and non-myopic collaborative coverage planning.
• The method effectively handles complex visibility constraints within an optimization framework.
• The approach is suitable for real-world applications like UAV-based infrastructure inspection.