Adaptive Nonlinear Model Predictive Horizon Using Deep Reinforcement Learning for Optimal Trajectory Planning

This paper presents an adaptive trajectory planning approach for nonlinear dynamical systems based on deep reinforcement learning (DRL). methodology is applied to the authors’ recently published optimization-based named model predictive horizon (NMPH). The resulting design, which we call ‘adaptive NMPH’, generates optimal trajectories autonomous vehicle system’s states and its environment. done by tuning NMPH’s parameters online using two different actor-critic DRL-based algorithms, deterministic policy gradient (DDPG) soft (SAC). Both NMPH variants are trained evaluated aerial drone inside a high-fidelity simulation results demonstrate curves, sample complexity, stability of adaptation scheme show superior performance relative our earlier designs.