Abstract:

The current longitudinal acceleration planning method based on prediction is complex and takes up a lot of hardware resources， so it is difficult to achieve mass production on low computing power controllers. Although the traditional planning method occupies fewer resources and has good real-time performance， it cannot guarantee the mass production requirements of high safety， comfort and reliability， and lacks the high versatility for adapting to multiple models. In order to solve the above problems， this paper proposes a longitudinal acceleration planning method. The constant speed cruise planning adopts the multi-dimensional optimization PID control method. With the help of the PID control idea， the error section and the time interval are reasonably divided， and the two-dimensional acceleration upper and lower limits are designed offline， which can be adapted to multi-vehicle and multi-duration configurations so as to improve the versatility of the algorithm. The car-following cruise acceleration planning adopts the model predictive control method based on dynamic prediction time domain， which predicts the vehicle motion state by considering the actuator efficiency and time delay， and then enhances the system safety. At the same time， the prediction time domain is dynamically managed to provide high versatility with multiple scenarios and multi-vehicle adaptation. And the solution complexity is reduced to meet the requirements of low resource occupation. Through the road test of a variety of mass-produced vehicles equipped with low computing power controllers， it is verified that the method has high safety and high reliability characteristics in the scenarios of constant-speed cruise and car following. After the10 000 km road test of the two mass-produced vehicles， 89.21 % and 86.95 % of the comfortable somatosensory ratio and statistical result for takeover less than one time per hundred kilometers show that the method meets the mass production requirements of comfort and robustness.

Key words: mass production, adaptive cruise, acceleration planning, dynamic time domain management, model predictive control