Highlights A human-like driving strategy implemented in automated vehicles may increase acceptance of these vehicles by their occupants and by other drivers on the road. Drivers’ speed selection in free-flow conditions on mountainous freeways can be characterized with three driving styles. The presented speed selection analysis helps quantitatively represent the behavior of a heterogeneous population of drivers. Continuous speed models developed for distinct driving styles help implement human-like speed control strategies in automated vehicles.
Abstract A driving strategy for autonomous vehicles (AVs) that is consistent with human behavior while demonstrating superior performance seems to have a good chance to be accepted by early AV users and be successful in the long run. Most of the past research focused on motion strategies affected by the presence of other vehicles. On the other hand, AV not constrained by other vehicles must select a safe and comfortable speed that is perceived as such by its occupants. This line of research is not well covered by the published work. The baseline speed, which is the speed AVs will follow without interaction with other vehicles, implemented via cruise control (CC) in modern vehicles is a constant speed consistent with speed limits and design speeds. A more advanced strategy of road-limiting speed control (RC) responds to influencing geometric features ahead of the AV’s current position. Neither of the two strategies considers AV occupants’ preferences. The current void in research is particularly obvious for free-flow conditions where baseline speeds must be implemented for extended periods of travel. Although the existing strategies have not been yet evaluated on roadways with demanding alignments and operating in free-flow conditions, the principles on which they are based provide a basis for skepticism if they can be acceptable to AV occupants. This study used the Tongji University driving simulator to evaluate the CC and RC strategies and their potential limitations in free-flow conditions on a mountainous freeway with complex alignments. Human speed-selection behavior was observed among a group of participating drivers. The clustering analysis of the data revealed three distinct driving styles: slow, fast, and consistent. The resulted analytical models provided human-focused road-dependent baseline speed profiles– a key element of the proposed human-like speed control (HC) strategy. The comparison of the existing speed-control strategies CC and RC with the proposed HC confirmed the limitations of the two existing ones if applied to roads with complex alignments. Considerable discrepancies were revealed between the baseline speeds produced with the existing and the proposed ones.
Towards human-like speed control in autonomous vehicles: A mountainous freeway case
2022-01-03
Article (Journal)
Electronic Resource
English
Speed Distribution Prediction of Freight Vehicles on Mountainous Freeway Using Deep Learning Methods
DOAJ | 2020
|A Speed Study of Mountainous Freeway Combined alignment
ASCE | 2014
|Research on Relationship between Operating Speed and Posted Speed Limit for Mountainous Freeway
British Library Conference Proceedings | 2010
|