On the challenges of jointly optimising robot morphology and control using a hierarchical optimisation scheme

Goff, Léni K. Le; Hart, Emma

Abstract

We investigate a hierarchical scheme for the joint optimisation of robot bodies and controllers in a complex morphological space. An evolutionary algorithm optimises body-plans while a separate learning algorithm is applied to each body generated to learn a controller. We investigate the interaction of these processes using a weak and then strong learning method. Results show that the weak learner leads to more body-plan diversity but that both learners cause premature convergence of body-plans to local optima. We conclude with suggestions as the framework might be adapted to address these issues in future.

Citation

Goff, L. K. L., & Hart, E. (2021, July). On the challenges of jointly optimising robot morphology and control using a hierarchical optimisation scheme. Presented at GECCO '21: Genetic and Evolutionary Computation Conference, Lille, France

• Presentation Conference Type: Conference Paper (Published)
• Conference Name: GECCO '21: Genetic and Evolutionary Computation Conference
• Start Date: Jul 10, 2021
• End Date: Jul 14, 2021
• Acceptance Date: Apr 26, 2021
• Online Publication Date: Jul 8, 2021
• Publication Date: Jul 7, 2021
• Deposit Date: Oct 13, 2021
• Publicly Available Date: Oct 13, 2021
• Publisher: Association for Computing Machinery (ACM)
• Pages: 1498-1502
• Book Title: GECCO '21: Proceedings of the Genetic and Evolutionary Computation Conference Companion
• DOI: https://doi.org/10.1145/3449726.3463156
• Keywords: learning, evolution, morphology, optimisation
• Public URL: http://researchrepository.napier.ac.uk/Output/2812272