As I have watched state DOTs over the last few years anticipate how to ready their systems for autonomous vehicles (AVs), it does not seem that they are getting much closer to prioritizing activities. How AVs will deploy and what type of roadways will be needed remain unknown outcomes of the biggest tech experiment for a century. Vehicle automation is approaching our roads in two ways: driverless ride hailing and Level 2/3 private vehicles.
Companies represented by GM, Ford, Waymo, Uber and Lyft have stated their plans to implement driverless services in urban settings in the next few years. Waymo is operating its Phoenix Early Rider vehicles with an empty driver seat and safety personnel riding in the back seat. It may only be a matter of months before Waymo introduces commercial driverless ride hailing in a cordoned area.
The second emerging segment of AVs is private vehicles. Included are both cars and trucks with the capability of off-loading driving duties from humans, but not operating in a driverless mode. Importantly, neither segment of these emerging AVs requires V2V or V2I technology. The vehicles are connected to the cloud via cellular connections that send and receive information.
But the vehicles utilize mapping and sensors to operate without communication to surrounding vehicles and infrastructure. So far the USDOT and states are spending much more effort on CV pilots than trying to understand the roadway environment challenges for AVs. Much of this response is probably due to the reluctance of industry to disclose the problems of early automated driving systems, leaving DOTs with no one to talk to about how to improve their roadways for AVs.
To launch its SuperCruise feature, GM made the decision to lidar map the entire limited-access highway system in the USA to create an operating domain for its vehicles. Where there is a lack of clarity about lane location or lane merging, GM disengages SuperCruise and requires the driver to resume control. If DOTs could work with GM to identify and understand these locations, perhaps improvement projects could better respond to the AV future, and even improve safety for all vehicles.
In San Francisco, GM is using its Cruise Anywhere employee ride service to test its automated driving system. As GM points out, the city is full of street conflicts that make AV operation difficult. Reports are that the Cruise vehicles have to avoid many streets in the city where the driving system cannot yet operate, adding significant time to a trip when compared with a human driver. Physical street challenges include faint traffic lights, complex roundabouts, narrow two-way streets, and short lane merges. Some of these urban street barriers could be eliminated or mitigated by DOTs, while others will eventually have to be handled by automated driving systems.
Opening a dialog between AV makers and roadway operators would yield useful information for both parties. But while we continue in a proprietary phase of technological development we are still some time away from the beginning of a meaningful exchange.
