Multi-Agent Path Generation Using Collision Time Intervals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for multi-agent path finding (MAPF) struggle with increased computational complexity and difficulty in generating paths in continuous time, especially when dealing with non-grid graphs and large numbers of agents, due to the need for extensive calculation to determine collision points.
Innovation Solution
A collision condition application device and method that extracts vertex-edge and edge-edge pairs where collisions are likely, calculates time intervals for collisions, and applies these conditions to a road map to generate paths efficiently in continuous time, using algorithms to identify potential collision points and time intervals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If continuous time path planning is applied to multi-agent on road map, then path generation accuracy is improved, but calculation complexity increases enormously
Solution Approach 1:
The continuous time path planning problem is segmented by identifying and extracting only the critical vertex-edge pairs and edge-edge pairs where collisions can occur. Instead of continuously checking all possible collision points, the method divides the road map into discrete critical pairs, reducing the continuous collision detection problem into a manageable set of discrete checks while maintaining accuracy.
Solution Approach 2:
The method performs preliminary action by pre-calculating and storing the critical vertex-edge pairs and edge-edge pairs before actual path planning. This pre-processing step identifies all potential collision scenarios in advance, allowing the path planning algorithm to efficiently check only these pre-identified critical pairs during execution, rather than performing exhaustive continuous collision detection.
2Quantity of substance
If the number of agents increases, then multi-agent path finding coverage is improved, but path generation time increases
Solution Approach 1:
The collision detection process is segmented into independent vertex-edge pair checks and edge-edge pair checks. This segmentation allows the system to process each pair independently and efficiently, reducing the computational burden when the number of agents increases. Each agent's path can be validated against pre-identified critical pairs without requiring exhaustive pairwise comparisons among all agents.
Solution Approach 2:
The method changes the parameter of collision detection from continuous spatial-temporal checking to discrete time interval checking based on agent speeds and distances. By parameterizing collision conditions in terms of specific time intervals and critical pairs, the system can efficiently handle multiple agents without exponentially increasing computational complexity.
3Reliability
If exhaustive collision detection is performed, then collision avoidance reliability is improved, but computational cost increases
Solution Approach 1:
The method extracts only the essential collision detection elements (vertex-edge pairs and edge-edge pairs) from the complete set of possible collision scenarios. By taking out and focusing only on the critical pairs where collisions can actually occur, the system maintains collision avoidance reliability while dramatically reducing the computational cost compared to exhaustive detection of all possible agent interactions.
Solution Approach 2:
The method performs partial collision detection by checking only the critical vertex-edge pairs and edge-edge pairs rather than all possible collision scenarios. This partial action is sufficient to ensure collision avoidance because these critical pairs represent all potential collision points, achieving reliable collision detection with reduced computational effort.
Data Source
AI summary
The present disclosure provides an extraction unit that extracts, in a case in which plural agents move on a road map including plural vertices and plural edges connecting the vertices, a vertex-edge pair and an edge-edge pair in which the agents are likely to collide with each other; a calculation unit that calculates a time interval in which a collision between both agents occurs; and an application unit that applies collision condition information.