Motion Planning using Reinforcement Learning for Electric Vehicle Battery optimization(EVBO)

The increasing demand for electric vehicle and autonomous vehicle as the alternate to the combustion-driven vehicle has motivated the research in the area of motion planning. Motion planmng is a complicated problem as it requires the consideration of multiple entities, mainly human behaviour. In this paper, reinforcement learning techniques are explored for the motion planning of an electnc vehicle(EV) while optimizing battery consumption. The EV travel time has also been evaluated under different reinforcement learning schemes. A traffic simulation network is developed for a high-traffic zone of Jaipur city using Simulation for Urban Mobility(SUMO) software. Model-based and model-free method like value-iteration and q-learning are applied to the developed traffic network. The results show that value iteration and q-learning have shown improved battery consumption. However, value iteration gives greater efficiency in terms of travel time as well as battery consumption.