Task Allocation and Motion Planning Strategies for Multi-robot Cooperation

The pick-and-place task is a common activity performed by robots in industrial settings. Traditional methods rely on fixed holders to maintain object positions, but this approach lacks flexibility and increases installation complexity and costs. To address this limitation, computer vision-based solutions are proposed to enable more flexible pick-and-place applications. Additionally, employing multiple robots in a shared workspace offers advantages such as increased performance and the ability to achieve complex objectives. However, coordinating and allocating tasks among multiple robots while ensuring collision avoidance remains a challenge. This study focuses on task allocation and motion planning, presenting two strategies: a predictive approach based on simulation optimization and a reactive approach based on control by multi-agent systems. The predictive approach utilises simulation optimization to anticipate collisions and optimizes task allocation, while the reactive approach emphasizes real-time decision-making and coordination among autonomous agents. The paper presents a case study for practical implementation on industrial robots.