Crash-Risk-Aware Integrated Predictive Control in Emergency Conditions for Intelligent Vehicles

Crash-avoidance systems in current intelligent vehicles (IVs) require accurate crash risk estimation and optimization of multiple control inputs. However, the uncertainties in vehicle states caused by the inherent noise in onboard sensors lead to increased errors in spatiotemporal crash risk calculations, which, in turn, affect the optimization of control inputs. To address these challenges, this article introduces a crash-risk-aware integrated predictive control (CRIPC), which computes crash positions and time-to-collision with uncertainties (UTTC) between vehicles under nonideal conditions by constructing an extensive elliptical vehicle geometry model to quantify the uncertainties arising from sensor noise. Additionally, CRIPC utilizes UTTC and the crash-point position between vehicles to dynamically adjust the crash-risk-aware repulsive field, which is subsequently used to define the objective function for CRIPC optimization. Validation results from real-vehicle experiments and driver-in-the-loop (DiL) simulations demonstrate that CRIPC effectively quantifies the motion uncertainties of surrounding vehicles, significantly enhancing crash avoidance capabilities while maintaining vehicle stability. Furthermore, CRIPC shows robustness to variations in road conditions and system uncertainties, meeting the real-time requirements of practical applications.