リモートセンシングとGISのジャーナル

概要

Comparison of four evolutionary algorithms for optimization of holding force in a climbing robot

Masike R and Janak Kumar B. Patel

Evolutionary algorithms (EAs) are stochastic search methods that mimic the natural biological evolution. In this work we provide an overview of four recent EAs and provide a framework for adoption in climbing robots. Four models based on EAs are introduced and compared for the optimization of the holding force in a Bernoulli holding pad for a climbing robot, by considering their important characteristics and their relevance to adaptive holding force. A brief code of each algorithm is presented to facilitate its implementation and use by researchers and practitioners. These EAs include the Differential Evolution (DE), MONEE implementation, Modified Genetic Algorithm (MGA) and the Memetic Algorithm (MA). The four EAs were applied to the popular MIT rule as objective function for the adaptive holding force, then to a real Bernoulli pad for climbing robot. MATLAB was employed for the rigorous comparison of the models in terms of the optimum solution obtained, the number of objective function evaluations corresponding to the optimum solution and the quality of the results. A statistical analysis was carried out and then an efficiency-rate metric was determined to assess the performance of each model. The results showed that the best performance came from a hybrid algorithm which incorporates desired characteristics for optimal holding force, thus a framework for adoption of EAs in climbing robots was developed.