Onboard Evolution of Understandable Swarm Behaviors. (23rd August 2019)

• Record Type: Journal Article
• Authors: Jones, Simon; Winfield, Alan F.; Hauert, Sabine; Studley, Matthew
• Is Part Of: Advanced intelligent systems. Volume 1:Number 6(2019)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2019-08-23
• Subjects: evolutionary robotics -- swarm robotics -- behavior trees -- explainability; Artificial intelligence -- Periodicals; Robotics -- Periodicals; Control theory -- Periodicals; 006.3
• DOI: 10.1002/aisy.201900031
• Languages: English
• ISSNs: 2640-4567

Abstract

Abstract : Designing the individual robot rules that give rise to desired emergent swarm behaviors is difficult. The common method of running evolutionary algorithms off‐line to automatically discover controllers in simulation suffers from two disadvantages: the generation of controllers is not situated in the swarm and so cannot be performed in the wild, and the evolved controllers are often opaque and hard to understand. A swarm of robots with considerable on‐board processing power is used to move the evolutionary process into the swarm, providing a potential route to continuously generating swarm behaviors adapted to the environments and tasks at hand. By making the evolved controllers human‐understandable using behavior trees, the controllers can be queried, explained, and even improved by a human user. A swarm system capable of evolving and executing fit controllers entirely onboard physical robots in less than 15 min is demonstrated. One of the evolved controllers is then analyzed to explain its functionality. With the insights gained, a significant performance improvement in the evolved controller is engineered. Abstract : A swarm of robots with high onboard processing power is used to run an evolutionary algorithm situated within the swarm. The swarm rapidly evolves fit behavior tree controllers, which can be automatically simplified and then understood and improved.