Road Friction Aware Adaptive Cruise Control using Robust Nonlinear Model Predictive Control with Uncertainty Quantification

Abstract

Road friction is a fundamental factor affecting vehicle safety, especially under adverse weather conditions such as rain, snow or ice. Reduced road grip increases stopping distances and the likelihood of loss of control, contributing to a significant number of traffic accidents each year. While modern Advanced Driver Assistance Systems (ADAS) are designed to enhance driving safety, many systems still lack the ability to adapt their behaviour in real time to varying road friction levels, limiting their effectiveness in low-friction or rapidly changing weather conditions. This thesis addresses this limitation by developing a friction-aware, curvature-adaptive Adaptive Cruise Control (ACC) framework based on a Robust Nonlinear Model Predictive Control (NMPC) formulation.

The controller anticipates road curvature and spatially-varying friction by incorporating predicted profiles into the optimization problem. To account for uncertainty, friction is represented through position-dependent bounds in the MPC, while stochastic realizations of friction are generated via Beta sampling and the lead vehicle’s acceleration varies randomly within feasible physical limits. These variations are reflected in dynamic safety constraints that adapt to evolving conditions ahead of the ego vehicle.

Soft constraint relaxation is introduced to maintain feasibility under conflicting demands such as sudden friction drops or aggressive lead vehicle deceleration. Despite the nonlinear nature of the problem and the presence of uncertainty, the controller operates using an efficient CasADi-based implementation in MATLAB.

Simulation results demonstrate that the proposed framework achieves robust, adaptive cruise control under friction and road curvature changes with stochastic disturbances. The integration of environmental awareness into predictive control enables safer, more responsive vehicle behaviour without sacrificing passenger comfort, and highlights the feasibility of embedding the friction information into future ADAS systems.