Automated Driving’s Journey: One Step Forward…
There’s advance – kind of – for ADAS, but high-level automation’s prospects haven’t been encouraging.
The AV sector was stunned by the late-October headline that the California Department of Motor Vehicles suspended robotaxi developer Cruise’s permits for deployment and testing of driverless vehicles. Except some weren’t stunned at all, as a drumbeat of criticism about Cruise vehicles’ sometimes erratic interactions with other road users intensified in the months leading up to the operational suspension.
Cruise’s difficulties in San Francisco came to a head in an accident in which a pedestrian was struck by a vehicle, then struck again by a driverless Cruise robotaxi. Cruise appeared to be less than forthcoming with details and data about the incident. When it subsequently was learned the Cruise vehicle that hit the pedestrian (who was seriously injured but not killed) in the aftermath didn’t “react” in a naturalistic fashion, prominent voices connected with AV development said it was time to take action.
Even proponents of the robotaxi use case conceded that all is not well in AV-development land if Cruise robotaxis’ performance in San Francisco represents the best one of the world’s most prominent and well-funded “full-stack” AV developers currently can deliver.
Is high-level automation the automotive industry’s Moby Dick, spurring a self-destructive pursuit that’s burned billions of dollars and produced scant payback in an ultimately misguided quest? I’d say it depends on your perspective.
Consider that Ford and Volkswagen in October 2022 pulled the plug on their jointly-funded AV developer Argo AI after a handful of low-impact pilot programs and years of optimistic talk about Argo’s progress. It’s no secret that Ford quietly absorbed some of Argo’s personnel and learning, but the monetary writedown for Ford and VW was enormous. I haven’t noticed VW say too much about high-level automation lately, as it and many other automakers and large suppliers largely have gone silent about “self-driving.”
Yes, some of the recent advances in ADAS can be attributed to the investments in high-level automation. But that’s only partly true. Although ADAS has become better and more affordable — and if nothing else, more widespread — ADAS still isn’t doing the job, either.
Mercedes-Benz’s new Drive Pilot ADAS was launched in two U.S. states in September 2023 and the company calls it an SAE Level 3 system, assuming full authority over both lateral and longitudinal vehicle control — with an important detail: the driver does not have to keep eyes on the road. Mercedes-Benz’s media demonstrations (in which SAE Media did not take part) encouraged behaviors such as playing a video game on the dashboard touchscreen while the system was engaged.
As there always will be with Level 3, there are caveats, chiefly addressing the range of conditions that must be met for Drive Pilot to be engaged. The system operates only up to 40 mph (64 km/h) and can be used only in “high-density” traffic. And more critically, those who’ve read the fine print believe the company still is evasive about who’s responsible in the problematic “handoff” period in which all Level 3 systems can “request” the driver to take control.
Instead of the quest to get drivers out of the equation, even conditionally, let’s get sensors and artificial intelligence and high-powered processors working on functionalities that might really make a difference. Forget “last-mile” – how about systems that help seniors or the physically challenged negotiate that last 40 feet into a narrow garage or a tight parking space? I want to press a button when approaching a tightly packed, unfamiliar parking structure and have automation drive my vehicle into an open parking spot. These specific capabilities have been demonstrated and effectively are child’s play for today’s technology. We need to take a break from Moby Dick and pursue mobility problems we can fix.