A Hybrid Supervisory Control System for Adaptive Co-Pilot Intervention in Drivers Education Training

open_access
Open Access
Author:
Kaboly, Nathaniel
Area of Honors:
Electrical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Azeemuddin Syed, Thesis Supervisor
Sudarshan Rao Nelatury, Thesis Honors Advisor
Keywords:
Hybrid Supervisory Control
Adaptive MPC
Driver Education
Shared Control
Stateflow
highD Dataset
MATLAB Simulink
Abstract:
Young drivers aged 15 to 20 are disproportionately represented in fatal and general crash statistics [1]. This highlights the need for automated safety override systems in driver education vehicles to prevent accidents during instruction. A critical challenge in developing autonomous driving systems is establishing a safe mechanism to assume control when a human driver makes a critical error. This thesis addresses that need by proposing and validating a novel Hybrid Supervisory Control Architecture (HSCA), implemented in a combined Simulink and Stateflow environment. The HSCA is specifically designed to assist human driving instructors by serving as a precise safety co-pilot, instantly overriding unsafe student commands, such as excessive speed or unintended lane departure, while maintaining smooth, predictable vehicle actions. The system’s intelligence is split between two components: an Adaptive Model Predictive Controller that continuously generates optimal steering commands for trajectory tracking, and a supervisory Stateflow chart that monitors calculated velocity and lateral position errors. When errors exceed defined safety thresholds, the Stateflow chart triggers autonomous intervention, utilizing time-delayed criteria to ensure stability before returning control to the student. The final architecture is tested against high-fidelity human trajectory data via the highD dataset to evaluate its robustness and ability to maintain safety parameters with dynamic, real-world data. Ultimately, this HSCA establishes a verifiable safety framework for driver education vehicles, allowing instructors to prioritize education over emergency intervention.