Development and Simulation of a Dual-Mode Autonomous Braking System for Human and AI Controlled Driving in an EV Formula Student Car
摘要
With many domestic vehicles now starting to incorporate autonomous features (ADAS) it is now a requirement for engineers to be aware of these elements and begin learning how to take advantage of the possibilities of their usage. Formula Student (FS) competitions around the globe are driving the development of young engineers by exposing them to autonomous vehicles by mandating that vehicles have autonomous capabilities with the German competition FSG needing ADAS capabilities to score maximum points. At Oxford Brookes the aim is compete at the highest level of FS which requires autonomous implementation into the successful human driven counterpart. The current focus is on the implementation of braking capabilities to the vehicle. This paper focuses on development and simulation of a braking system which delivers 1 g of deceleration during on track running. This involves development of a Simulink model, design of multiple PID and Fuzzy Logic Controllers and implementation of a braking model which is based on numerical methods. This methodology allows for quantitative performance data to be paired with control system performance to understand the real-world effects of slight variations within the control system. Research has found that a 2 degree of freedom PD controller provides the best performance with an absolute error value of 3.0498% which results in braking performance which closely matches the ideal performance with a difference in 0.1 m/s at the end of the braking event. The research also looked into the effects of Fuzzy Logic Controllers which shown that the performance required from the control system cannot be reached with the FLC’s. Tests showed significant instabilities when the performance was close to the set point and only showed acceptable stability with significant undershoot in the performance which is not acceptable for autonomous vehicle performance.