A strategic search algorithm in multi-agent and multiple target environment

Afzalov, Azizkhon; Lotfi, Ahmad; Aydin, Mehmet Emin

Abstract

The aim of this study is to investigate how to solve the path-planning problem of multiple competing players towards moving targets within a dynamically changing environment. A novel approach is needed to exceed the classical solutions for single or multiple agent search algorithms. An assignment strategy is introduced for pursuing agents. They can compute the distance towards the targets, while all players are at the current state. The reason to use the assignment strategy is to find a better solution, especially in the round circle cases, where pursuing agents trap and outmanoeuvre the targets. This study set out to explore how the agents can outsmart targets by exploiting repeated A* searches. To find a solution to such a dynamic problem, a heuristic search algorithm Strategy Multiple Target A* (STMTA*) will be developed and implemented. These are promising AI approaches. Pursuing agents can freely communicate and share information. A multi-agent framework will be used. The experimental results suggested that STMTA* is quicker in number of steps and successful in respect of catching targets on each test run approximately by 27%.

• Presentation Conference Type: Conference Paper (published)
• Start Date: Dec 11, 2020
• End Date: Dec 13, 2020
• Acceptance Date: Nov 30, 2020
• Online Publication Date: Aug 5, 2021
• Publication Date: 2021
• Deposit Date: Sep 14, 2021
• Pages: 195-204
• Series Title: Lecture Notes in Mechanical Engineering
• Book Title: RiTA 2020
• ISBN: 9789811648021
• DOI: https://doi.org/10.1007/978-981-16-4803-8_21
• Public URL: https://uwe-repository.worktribe.com/output/7757940