Multi-Agent Search with Interim Positive Information Haye Lau, Shoudong Huang and Gamini Dissanayake ARC Centre of Excellence for Autonomous Systems (CAS) University of Technology, Sydney NSW, Australia {hlau, sdhuang, gdissa}@eng.uts.edu.au Abstract – A problem of searching with multiple searchers and scouts is presented. Unlike most search problems that terminate as soon as the target is found, successful detections by scouts only improve on the current knowledge of the moving target’s location, such that the searchers can more effectively find and service the target in the future. The team must correspondingly plan not only to maximize the probability of the searchers directly finding the target, but also give them the best chance of exploiting any new information from potential scout detections. It is shown that this need to plan for replanning can be addressed by equivalently solving a series of simpler detection search problems that always do terminate on detection. Optimal and heuristic solution methods for this Searcher/Scout problem are derived, such that the capabilities of all the sensing platforms in a search task are harnessed even when only a subset are capable of actually servicing the target. Index Terms – Multiple agent search, target search, branch and bound, optimal searcher path problem. I. INTRODUCTION Optimizing the search for a moving target constitutes an important problem with applications in many rescue and security scenarios. Given all available knowledge of the environment structure, the likely target locations and searcher capabilities, the objective in general is to find the best trajectory for the searcher to travel such that the chance of finding the target is maximized [1]. Using multiple agents can clearly further improve the chances of finding the target within a limited amount of time [2][3]. Although many applications involve agents that are all capable of servicing (e.g. rescuing or engaging) the target, there may be situations where some of the agents lack the equipment or skill to do so [4]. Those agents should then do what they can to share the workload of the more capable searchers. This paper considers the search of an area with a heterogeneous team in which the searchers are aided by mobile sensor scouts that can provide target information but cannot directly service the target. Envisaged scenarios include a team of fire-fighters entering a burning building with a number of scouting robots, which then collectively search for the moving victim until a fire-fighter can arrive to render aid. Unlike standard detection search problems that terminate as soon as the target is found, the target is free to continue moving after being detected by a scout and may disappear again from view. Scout detections therefore serve to improve the available target information on which future search actions can be based. Given this opportunity for the agents to react to positive target information, they should thus balance between maximising the searchers’ own ability to directly find the target, and ensuring that they can fully exploit potential detections by the scouts. It is shown that this need to plan in anticipation of replanning can be addressed by equivalently solving a series of simpler detection search problems that always do terminate on detection. Building on our recently developed optimal solution for a related detection search problem [1], optimal solution methods for the above Searcher/Scout problem are derived. Given the high computational complexity of optimally solving the problem, a number of faster heuristic methods are also proposed and compared with the optimal approach. This paper is organized as follows. Section II describes the multi-agent search problem addressed in this paper and provides a detailed definition. A strategy for obtaining optimal policies is presented in Section III and a number of heuristic methods are proposed in Section IV. Section V illustrates the optimal and heuristic solutions with the aid of an example. Section VI discusses the related work and Section VII summarises the paper. II. PROBLEM DESCRIPTION A. Searching with the aid of scouts --- Overview Figure 1. Searcher examining an area with the aid of scouts Figure 1 depicts a typical scenario that motivates the problem considered in this paper. Two types of agents, searchers (e.g. humans/capable robots) and scouts (mobile sensors), are tasked with locating a target moving through an environment divided into a set of cells. Each agent may move from one cell to another at each time step, subject to the environment’s structure. A searcher can detect the target with some finite probability if both it and the target occupy the