Design and Test of Speed Tracking Control for the Self-Driving Lincoln MKZ Platform

To accelerate self-driving technology developments, testing platforms modified from production passenger cars are widely used nowadays, but the powertrain models and control access to engine/motor are often not available, e.g., the popular hybrid Lincoln MKZ platform only opens brake/throttle pedal control protocol for longitudinal motion. These limitations motivate us to model the powertrain dynamics by experiments and to design new speed controls, differing from the conventional cruise control that manipulates engine/brake torque directly with accurate models. This article explores the powertrain modeling and speed control algorithm design of these testing platforms, by taking the MKZ as an example. Two computationally-efficient algorithms are designed, i.e., a PID+C control when future information (i.e., reference speed and road slope profiles) is unavailable, and a preview control if the information is provided a priori. To guarantee speed tracking errors within a given safe range, a barrier control is also designed to supervise the two controllers, which adds additional brake/throttle actions when the error is approaching the boundary. The algorithms are implemented and tested on the Mcity MKZ platform, experimental results show the speed-tracking performance and the bounded errors achieved by the barrier function.