Controlling car braking by means of electromechanical actuation on the pedal

As the autonomous driving paradigm started to propel the automotive field, the need to automate steering, acceleration and deceleration, the fundamental actions used to drive a car, arose. This thesis focuses on the development of an autonomous braking control that must be stable and effective in the real world, which is, due to the high complexity of the manoeuvre and the systems used to operate it, an ever uncompleted topic. In the automotive sector the most used solutions are brake-by-wire systems, where electrically controlled actuators are generally used to control basic car functions. Even if practical it’s cumbersome to reach a good compromise between performance and costs with this approach. Together with Bylogix s.r.l. we have therefore developed a solution to actuate the legacy, preinstalled braking system of a car directly operating its braking pedal with an electric DC motor and leveraging a specifically designed mechanical system. This solution is purposely designed for the experimental autonomous vehicle the company is developing. My study went through the whole remotely automated braking system consisting of a DC motor, external mechanical actuator, brake pedal, brake booster, master cylinder, hydraulic circuit and brake disc, and ended up producing a physical model for the best estimate of the behaviour, stability and the related requirements of the brake controller. Given the system characteristics, considering the goals and the need to make the control suitable also for the real world, I decided to use a "Hybrid Gain Scheduling" in which two PID controllers are implemented. The first is used for “normal braking”, that is lowering the speed down to the input target speed selected by the autonomous driving pilot with possibly the smoothest deceleration. The second is used for “emergency braking”, meaning exceptionally alarming situations with high risk to loose the car’s control, when a hard braking is required to prevent harmful conditions. In this case the input is the optimal slip of the wheel to obtain the best deceleration without loosing the drivability, like wheels blocked and skidding. The switching between the two controllers is done by a proper "switch logic” where it is needed to have a comparator with hysteresis and to know the optimal value of the slip, that depends on the road/tyre condition. A model of the whole braking system has been implemented in Matlab and Simulink, allowing to tune the PIDs coefficients and to evaluate the response of the designed system for different braking scenarios and tyre/road friction conditions, in terms of robustness and performance.