MOGA 4WD: multi-objective genetic algorithm for four-wheel drive electrical vehicle torque distribution in challenging conditions

Abstract

This article aims to present a strategy for multi-objective optimization based on torque distribution for electrical 4WD (four-wheel drive) vehicles. By considering applications on uneven terrain, common to the navigation of tractors, off-road vehicles, or even mobile robots, an algorithm is developed having as input the vehicle attitude and output the controlled torque on each actuated wheel. The main criterion adopted is to guarantee the execution of a stable trajectory. And, to avoid wheel slippage, which occurs when low torques are applied, as well as vehicle rollover, which can occur in the presence of high torques, it is necessary to use two objective functions. To find the Pareto optimal solutions, the simplified dynamic model of a vehicle is adopted, considering a quasi-static motion. For each vehicle, its electrical, mechanical, and geometric characteristics can be used as formulation constraints. From an optimization performed offline, and adopting a polynomial approximation-based approach for real-time application, simulations and experiments show an interesting behavior: solutions that go beyond allowing the ascent of simple ramps or the overcoming of smooth obstacles are found - it is possible, for example, to climb ramps with high slopes, taking the vehicle to the limit between stability and instability.