Development framework for a racing-oriented torque vectoring algorithm and controller proposal

Laburpena

Electrification has drastically changed the way road car are designed. The floor of the vehicle chassis has substantially changed to contain the batteries, and electric motors can be placed practically inline to each axle. Furthermore, the compactness of the electric motors permits a layout where each wheel is powered by a single motor. Independently powering each vehicle’s wheel brings several advantages in terms of manoeuvrability and stability. Torque vectoring refers to the algorithm that utilizes these independent motors to alter the yaw of the vehicle for the aforementioned purpose. In this PhD dissertation, a torque vectoring algorithm is designed for a race car, following the V-shape development framework, widely used in the automotive industry. Race cars differ in many aspects from road cars, they generally have stiffer suspension, lower centre of gravity, grippy tyres, etc. They also feature a more neutral behaviour compared to the generally understeering behaviour or road cars. In this way, the front axle can generate higher lateral force, vehicle may increase its lateral acceleration, and therefore the car can negotiate faster the corners. However, this decrement on the understeering gradient may also suppose a decrease on its stability margin, so the car is trickier to drive. In this work, the torque vectoring is applied on a vehicle which features a close-to-neutral balance (therefore a with a little margin to increase its steady state lateral acceleration), aiming to keep this balance most of the times and preventing the vehicle from becoming unstable for the driver. Besides, since the algorithm needs to work in harmony with the driver -especially given the fact that the vehicle will presumably be driven at the limit handling area- the algorithm is designed and tested in conjunction to a human driver using a Human-In-the-Loop system.