A Conflict-Based Search Framework for Multiobjective Multiagent Path Finding

Automated routing of droplets for DNA storage on a digital microfluidics platform

“…Conflict-based searching might be particularly efficient if the problem is formulated in terms of multiagent path finding. 62–64…”

Distributed mean-square consensus of heterogeneous multi-agent systems with packet losses

“…In the middle of this spectrum lies the dynamically-coupled methods such as M* [26] and CBS [20], which begin by planning for each agent a shortest path from the start to the goal ignoring agentagent collision and then couple agents for planning only when needed to avoid collision. Dynamically-coupled methods are improved and extended in many ways [7], [11], [16], [18]. All these MAPF planners find start-goal paths without considering any intermediate target locations.…”

“…Although a few approaches have been developed [3], [7], [12], [17], [19], [24], [28] to handle MCPF and its variants over the past few years, most of them ignores or simplifies the task duration at a target location, which is ubiquitous in practice and is the main focus of this paper. In other words, when a robot reaches a target to execute the task there, it takes time for the robot to finish the task, and during this period, the robot has to occupy that target location and thus blocks the paths of other agents.…”

“…The first method has no solution optimality guarantee. It begins by ignoring the task duration and using an existing planner for MCPF (such as [19]) to find a set of paths, and then post-processes the paths to incorporate the task duration while avoiding collision between the agents. Here, the postprocessing leverages the idea in [7] to build a temporal planning graph (TPG) that captures the precedence requirement between the waypoints along the agents' paths, and then add the task duration to the target locations while maintaining the precedence requirement to avoid agent-agent collision.…”

“…To find an optimal solution for MCPF-D, we develop our second method called Conflict-Based Steiner Search with Task Duration (CBSS-D), which is similar to CBSS [19] by interleaving target sequencing and path planning. However, CBSS-D needs to consider task duration during planning, and there are two main differences between CBSS-D and CBSS.…”