Hu, Haoqi (Tongji University), Pan, Wei (Tongji University), Liu, Hanghang (Tongji University), Zhang, Lin (Tongji University), Chen, Hong (Tongji University)

Automated Hyperparameter Tuning of NMPC Based on Differentiable PMP for Electric Vehicles

Scheduled for presentation during the Video Session "On-Demand Video Presentations" (VP-VP), Saturday, November 22, 2025, 08:00−18:00, On-Demand Platform

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on April 2, 2026

Abstract

Extreme driving maneuvers involving sharp turns, aggressive acceleration, or abrupt deceleration on low-friction surfaces can induce rapid vehicle state variations and potential stability loss, making active safety control crucial for autonomous vehicle operation. The development of effective control strategies proves particularly challenging for multi-actuator systems with multiple constraints, where conventional parameter tuning approaches require substantial expert knowledge and often yield suboptimal performance. To address these limitations, this study presents an automatically-tuned torque vectoring control (TVC) strategy to enhance vehicle dynamic performance. The approach involves establishing a nonlinear vehicle model and designing a model predictive controller (MPC) that computes optimal yaw moments by tracking reference yaw rate and sideslip angle. The resulting nonlinear programming problem is solved using Pontryagin's minimum principle (PMP), with a novel differentiable PMP developed to compute state trajectory and control input gradients with respect to MPC weight parameters, thereby enabling gradient-based parameter adaptation. Comprehensive hardware-in-the-loop experimental validation under standardized double lane change condition demonstrates that the proposed self-tuning method significantly enhances controller performance, vehicle maneuverability, and stability compared to manually-tuned TVC strategies.