Multi-Agent Path Execution with Uncertainty

Opportunity

Unexpected events occur in many real-world multi-agent systems and applications, but they are usually not managed well. Such events can downgrade the effectiveness of path execution. Hence, there is a need for safe and efficient execution of multi-agent path plans considering arbitrary unexpected delays on the movements of the agents.

Technology

We have studied the multi-agent path planning and execution with uncertainty problems and developed a solution framework. The aim is to coordinate multiple agent movements in real-time to avoid collisions/deadlocks and maximise the number of agents moving concurrently to reach their goal locations effectively. The main idea of our framework is to unblock a maximum subset of the agents in each timestep while keeping the residual path plan feasible.

Figure 1: A screenshot of the multi-agent path execution demo. 

Figure 2: A block diagram on the high-level design of the method. 

Applications & Advantages

• Our framework is versatile, applicable to diverse multi-agent scenarios.
• Ideal for situations where unforeseen delays complicate path planning.
• Especially useful in dynamic environments where path execution adjustments are necessary.
• Examples include robot planning in airports or on streets, where challenges arise due to unpredictability of obstacles like pedestrians.