期刊名称：Journal of Automation, Mobile Robotics & Intelligent Systems (JAMRIS)
印刷版ISSN：1897-8649
电子版ISSN：2080-2145
出版年度：2019
卷号：13
期号：1
页码：15-19
DOI：10.14313/JAMRIS_1-2019/2
出版社：Industrial Research Inst. for Automation and Measurements, Warsaw
摘要：In order to be fully autonomous, robots have to be resilientso that they can recover from damages and operatefor a long period of time with no human assistance. Tobe resilient, existing approaches propose to change therobots’ behavior using a different control system whena hardware fault or damage occurs. These approachesare used for robots which have fixed morphologies.However, we cannot assume which morphology wouldbe optimal for a given problem and which morphologyallows resilience. In the present paper, we introducea new approach that generates resilient artificial modularrobots by evolving the robot morphology along withits controller. We used a multi-objective evolutionaryalgorithm to optimize two objectives at a time, whichare the traveled distance of a damage-free robot and thetraveled distance of the same robot with damaged parts.The result of preliminary experiments demonstrates thatduring evaluation, when robots are deliberately faced tomotor failures, the evolution process can optimize andgenerate new morphologies for which the robot’s behavioris less affected by damage. This makes the robot capableto recover its ability to move forward.
