TempGA: A Temperature-Inspired Adaptive Genetic Algorithm for Solving 7DOF Inverse Kinematics Problems

Abstract

Temperature fluctuations can influence the stability of Deoxyribonucleic Acid (DNA), affecting recombination and mutation rates. Inspired by these biological mechanisms, we introduce a new algorithm named a temperature-based adaptive Genetic Algorithm (TempGA). Our primary objective is to enhance Genetic Algorithm’s performance by incorporating biological temperature effects, improving the balance between exploration and exploitation. By integrating temperature-sensitive mutation and crossover variations, along with the concept of DNA hot and cold spots, our approach dynamically adjusts genetic operators based on population diversity and evolutionary progress. TempGA employs a temperature adaptation strategy, where mutation and crossover rates follow a function that increases with temperature. It selectively applies genetic operators depending on the algorithm’s exploration-exploitation phase, ensuring efficient convergence. We evaluated two variations of our algorithm, RTempGA (real-valued representation) and BTempGA (binary representation) on an Inverse Kinematics (IK) problem with collision constraint, on a 7-Degree-Of-Freedom (DOF) robotic arm. Our algorithms were compared against standard Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Simulated Annealing (SA), and Differential Evolution (DE). Our experimental results demonstrate that BTempGA outperforms bio-inspired alternatives, achieving higher precision in solving IK problems by consistently achieving low fitness values across all target points. This suggests that our approach is well-suited for real-time applications in robotics and animation, enabling more responsive, adaptive virtual agents with improved motion accuracy and interactive capabilities.