Title: Empirically Comparing Three Multi-Objective Optimization Approaches for the Automated Evolution of Snake-Like Modular Robots

Year of Publication: Nov - 2014
Page Numbers: 175-183
Authors: Wei Shun Chee , Jason Teo
Conference Name: The International Conference on Artificial Intelligence and Pattern Recognition (AIPR2014)
- Malaysia

Abstract:


This paper explores the use of multi-objective evolutionary algorithm to automatically design and optimize heterogeneous snake-like modular robot through artificial evolutionary process by taking consideration of two contradiction objectives which is to maximize the modular robot forward moving behaviour and minimize the complexity of the snake-modular robot. A hybridized Genetic Programming and self-adaptive Differential Evolution algorithm is implemented in this work to simultaneously co-evolve both the morphology and controller of the robot in the artificial evolutionary process. Several experiments had been conducted in this work by using different MOEA approaches. It was found out that the post MOEA approach is able to obtain better Pareto-optimal front solutions where the snake-like modular robot evolved using this method are more diverse and having higher performance score. Interesting findings had been obtained in this work showing that the snake-like modular robot is likely to have better moving behaviour by having higher number of segments. However, the performance of the snake-like modular robot will start decreasing if excess number of segments is being used. In conclusion, promising results had been obtained in this work by implementing the MOEA to automatically design and co-evolve heterogeneous snake-like modular for the forward moving behaviour.