Automated Vehicle Field Testing Under Various Weather Scenarios Conducted in Ohio Provided Lessons Learned Based on Eight Driving Scenarios.
East Liberty, Ohio, United States
Automated Vehicles and Adverse Weather Phase 3 – Final Report
Summary Information
Federal Highway Administration (FHWA) Road Weather Management Program has been exploring Automated Vehicles’ (AV) potential shortcomings during adverse weather conditions. The Automated Vehicles and Adverse Weather Phase 3 (AVAW3) explored how adverse weather and road conditions in different driving environments affected AV dynamics and operations, driver behavior, communications, and AV sensor capabilities through field testing funded by the FHWA. The study used AVs with the Society of Automotive Engineers (SAE) Level Two (partial automation) and Level Three (conditional automation) features. The study considered various driving scenarios and weather conditions for the field tests conducted in 2020 through 2021 at a testing facility in East Liberty, Ohio. The first round of field tests took place in Spring/Summer road weather conditions and evaluated the performance of the AVs in four different driving scenarios: (i) work zone lane change with barrels; (ii) work zone lane closure with lane markings; (iii) pavement markings with brake marks; and (iv) pavement markings with disappearing shoulder. The second round of field tests were conducted under winter road conditions and evaluated the performance of AVs in four different driving scenarios: (i) lane keeping; (ii) right lane changing; (iii) movements during green light at a signalized intersection; and (iv) stopped car detection.
- Equip the vehicles with redundant driving automation system components for enhanced safety in adverse weather conditions. In this study, during certain adverse weather conditions, the test vehicles lost localization, steering control, and veered off the desired paths. Extra or redundant automation components may be needed to function as backups and ensure safety at all times and weather conditions.
- Consider using LiDAR-based perception over camera-based one under winter weather conditions. This study found that the driving capability of AVs with camera-based perception system was not reliable, LiDAR, however, performed well under winter weather conditions, though with limited capabilities in perceiving some roadway conditions including pavement marking detection.
- Be aware that a high-performing AV under good weather conditions might lead to driver over-confidence in AV performance during inclement weather. This study demonstrated that such over-reliance on AV performing impeccably under all weather conditions may lead to distracted driving, disengagement, and an inappropriate use of automation.
- Testing multiple sensors can enable further insight into AV performance during field testing. This study suggested that the testing of multiple brands of advanced sensors could enable researchers to gain a deeper understanding of the reasoning behind vehicle performance under adverse weather conditions, as well as different driving environments.
- Examine the effect of critical driving conditions on steering torque. The study noted the importance of capturing the impact of critical driving conditions (e.g., sharp maneuvers or sudden obstructions) on the vehicles' steering torque, which could be addressed by developing scenarios with higher variances in steering torque.
- Examine the performance of AV control and perception systems with varying road configurations. This study recommended testing vehicles with SAE Level three or higher levels of automation by modifying the road configurations, such as creating challenges involving limited or obstructed field of view.
- Consider the driver as a fallback-ready user. It is important to understand driver behavior in different weather conditions to enable the development of effective AV control algorithms.
