Tesla Full Self-Driving Goes Beta: Who Needs DMS?

So pervasive is the belief that Tesla is but one simple over-the-air update away from self-driving utopia that technology developments in vision-based driver monitoring systems (DMS) go largely unnoticed. Who needs DMS and monitoring driver engagement when Tesla is in the home stretch on its way to “Full Self-Driving”?

Not so fast. Tesla may be getting ahead of itself given the results of a recent study from Consumer Reports, headlined “Cadillac’s Super Cruise Outperforms Other Driving Assistance Systems” which concluded “Other automakers close in on Tesla’s Autopilot, now a distant second, in Consumer Reports’ new ratings of 17 systems.”

Kelly Funkhouser, head of connected and automated vehicle testing for Consumer Reports (CR) commented, “Even with new systems from many different automakers, Super Cruise still comes out on top due to the infrared camera ensuring the driver’s eyes are looking toward the roadway.”

CR’s conclusions echo those from Euro NCAP (New Car Assessment Program) and UK-based Thatcham Research, which recently co-published the “Assisted Driving Grading” assessments, as well as the assessment of Missy Cummings, a professor at Duke University and director of Duke’s Humans and Autonomy Laboratory, in her report called “Assessing Tesla Model 3s’ Autopilot Interactions with the Driver Monitoring System,” presented in this video.

The message from CR, Euro NCAP, Thatcham Research, the National Transportation Safety Board (NTSB)  and Missy Cummings could not be clearer: Dial back the hoopla, Elon, and install proper DMS.

For a market which barely exists today, there is an extraordinary level of technology innovation and product development taking place in automotive vision-based DMS. This mostly goes unreported because the leading vendors are practically unknown companies, such as Eyesight Technologies (Herzliya, Israel), Jungo Connectivity (Netanya, Israel), Seeing Machines (Canberra, Australia), Smart Eye (Gothenburg, Sweden) and Xperi (San Jose, CA, US).

Although largely overlooked by the mainstream media, the safety benefits of DMS have not gone unnoticed by advisory and regulatory bodies in many regions around the world, in particular in China, Europe and the U.S. The developments in these regions can be summarized as follows:

China
Almost all automotive DMS design-win activity that I have tracked in 2020 looks to be related to China. As the country with the second highest number of traffic-related deaths each year (about 255,000 according to figures from the WHO), China has already introduced a legislative roadmap to deploy DMS to detect events such as driver cellphone use, eating and smoking, and to enforce seatbelt use for all occupants.

This video from Eyesight Technologies provides an excellent overview and shows a practical application of vision-based DMS marrying computer vision processing with probabilistic AI to bring demonstrable safety benefits to roadways in China.

Europe
European legislation for vision-based DMS is several years behind China, but my interpretation of developments is that the complexity and sophistication of DMS will be vastly superior in Europe than in China.

In particular Europe looks to be leaning towards “driver state monitoring.” European researchers are combining human factors science and behavioral research to develop high-level DMS signals for attention, distraction, drowsiness, engagement and impairment, which provide a real-time understanding of the driver’s cognitive state as it applies to accident risk.

That’s an elaborate way of saying that all the potential alerts, beeps and chimes from the ADAS systems can be almost completely eliminated provided the driver remains attentive to the roadway and engaged in the driving task. “Defining Safe Automated Driving” published by Thatcham Research is an excellent resource clearly sign-posting the way ahead for driver state monitoring policy in Europe and I encourage you to read it in full.

The European Parliament has mandated the introduction of vision-based DMS in all vehicles with four or more wheels (cars, trucks, vans, buses, coaches) by updating the European General Safety Regulation (GSR) for the type-approval requirements for motor vehicles.

At 106 pages, the full policy document is a complete snore-fest and an excellent cure for insomnia. My interpretation is that the legislation as it relates to vision-based DMS (presented on page 13 of Annex II) will be introduced in two stages:

• From June 2022 to June 2024 for driver availability monitoring (vision-based DMS for Level 2+ and Level 3 vehicles).
• From June 2024 to June 2026 for advanced driver distraction warning systems (vision-based DMS for Level 0-2 vehicles).

Independently from the European Parliament and the updated GSR, Euro NCAP is also set to include vision-based DMS testing as part of its five-star crash rating program from 2024. This was originally scheduled to commence in the 2022 tests and the two-year delay is the result of complications related to Covid-19.

Although Euro NCAP has yet to publish specific details of its testing protocols for vision-based DMS, product developments in the market suggest an emphasis on the rearview mirror for location of the infrared components and a trend towards all-cabin occupant monitoring systems (OMS) to enforce seatbelt use.

What is clear is that much of the work undertaken by Thatcham Research and Euro NCAP for the definition of detailed vision-based DMS testing protocols will be adopted by advisory bodies in other regions in quick succession.

United States
I wrote in July about the legislative position for DMS adoption in the United States. Little has changed since then, with the Moving Forward Act (H.R.2) having passed the House of Representatives and moved on to the Senate.

As discussed in EE Times in a piece entitled “Lack of ADAS Benchmarks Is Haunting Car Industry” , I highlighted that NHTSA could elect to improve road safety through changes to the NCAP assessment ratings in the U.S. This is most likely to happen in collaboration with Thatcham Research and Euro NCAP.

According to a May 2018 report in the Wall Street Journal, Tesla rejected integrating vision-based DMS with Autopilot because it deemed the technology ineffective. Lawmakers and regulators in Europe wholeheartedly disagree with Tesla’s assessment.

The critical piece of information to take from this article is that, irrespective of the capabilities of Autopilot and “Full Self-Driving,” beyond June 2026 Tesla will not be permitted to sell a single vehicle in Europe without a suitable vision-based DMS. So with self-driving supposedly on the home stretch, who needs DMS? The answer is simple: Any automaker that wants to sell in Europe.

William Wallace, manager of safety policy for Consumer Reports summed up: “It’s critical for active driving assistance systems to come with safety features that actually verify drivers are paying attention and are ready to take action at all times. Otherwise, these systems’ safety risks could end up outweighing their benefits.”

The post Tesla Full Self-Driving Goes Beta: Who Needs DMS? appeared first on EE Times Asia.

from EE Times Asia https://ift.tt/366Qza9