Predictive Modeling of a Flexible Robotic Arm using Cohort Intelligence Socio-Inspired Optimization

Socio-inspired metaheuristics can accelerate industry 4.0 implementation. In the current work, cohort intelligence socio-inspired metaheuristic has been employed for predictive modeling of a flexible robotic arm system. Robotic arm acceleration was modeled based on robot structure reaction torque as the input parameter. This model had a second order transfer function structure consisting of one zero and two poles for better predictability of the robotic arm dynamics. The model parameters were estimated using Cohort Intelligence (CI) socio-inspired algorithm. For comparisons, model parameters were also estimated by a nature inspired method - Genetic Algorithm (GA). Estimation results indicate that both CI and GA obtained similar FIT of system dynamics; 74.78 % and 74.27 % respectively. However, due to the higher number of the average function counts in GA (5480) as compared to CI (2379), the computation time for the best cost function was found to be much lesser in CI (10.65 seconds) as compared to GA (45.77 seconds). Thus, for similar fitting predictive models, the CI converges to the optimum model parameters much faster than GA. This indicates the superiority of the socio-inspired CI metahueristic towards faster predictive modeling of complex robotic systems in industry.