Stiffness-Aware Genetic Algorithm for Robotic Path Finding Optimization

This work presents a genetic algorithm to evaluate a suitable path for robotic manipulation of elastic objects. The goal is to find a path, given an initial and final position, that accounts for the distance covered by the robot while minimising the forces perceived. These forces are generated by flexible objects that are difficult to model analytically. The proposed model-free approach considers these elastic forces in the fitness functions, making the genetic algorithm stiffness-aware. Furthermore, a dynamic exploration strategy allows for the convergence to effective solutions in a finite number of iterations. A simulated analysis of the suitable parameter configuration is performed for the robotic platform employed in the experiments. These parameters are then used in the validation of the method, demonstrating positive outcomes of the proposed approach in terms of convergence and effectiveness.