AV Regulation Crucial to Consumer Acceptance

The auto industry is fast approaching a regulatory crossroads that could waylay its trip towards producing and selling fully autonomous vehicles.

The lack of standardization, especially regarding safety, is becoming more problematic as recently highlighted by the United Nations whose officials acknowledge autonomous vehicles could “save billions of dollars every year and help combat climate change while reducing congestion and emissions”. However, these advancements can only be realized on a global scale if countries work together on the necessary international laws and regulations.

Otherwise, with different systems in place in different countries, all autonomous vehicles would stop at the border, meaning that international transport of people and goods could not benefit from these technologies. Furthermore, manufacturers would have to develop different vehicles for every country, making them prohibitively costly.

The lack of regulatory harmonization may also prolong the process of getting to fully self-driving vehicles, making them more expensive, as well as delaying homologation across international borders and even between state lines in the US. “The collaboration the UN is pushing, to create something consistent, is a great step forward,” notes Elaina Farnsworth, CEO of the Next Education, which helps train workers for careers in intelligent transportation and mobility. The company counts among its clients automakers, insurers, suppliers, as well as state and local governments.

Farnsworth said the lack of safety and operating standards among different US states creates a difficult situation even within one country: “It’s maddening for OEMs, I’d imagine, to navigate those statewide regulations and global standards.”

The current bifurcation of regulation by the vehicle on the federal level and drivers at the state level is one key hurdle. Michael Clamann, senior human factors engineer and research associate at the University of North Carolina Highway Safety Research Center, sums up the situation concisely: “There’s no agreed upon measure to determine whether an AV is safe enough. This is partially true for conventional vehicles, too. We collect stats, make changes to roadways but our benchmark is to test teenagers and not do anything else after that, unless there’s a change in (drivers’) vision.”

Clamann then added: “What we’re missing from AVs right now is a driver’s license test. NHTSA has requirements for safety, the crash-worthiness of vehicles, but not standards for how automation will work, effectively the driver. Someone at the federal level will have to come up with a means of determining whether the driving AI (in AVs) is safe and OK to be on the road.” While the American public has shown skepticism about AVs in some polls, a recent survey by Partners for Automated Vehicle Education (PAVE) found wary respondents would be more open to self-driving vehicles through more education and firsthand experience with them.

Safety first

Importantly, safety was the top choice as most compelling case for why AVs could be a change agent for a better society. More than one third of responses backed safety, with nearly half (45%) of those polled saying AVs can save lives, compared to just 19% disagreeing. That still left more than a third unsure.

Industry insiders and the general public would seem to agree that AVs will ultimately be safer than human-driven vehicles but, given the lack of benchmarks, just how much safer will they be and when? “In general, AVs will be safer but how safe is still an open question,” noted Gary Streelman, director advanced engineering concepts for Tier 1 supplier Marelli. Streelman cited government statistics estimating “75% to 80% of accidents are driver error”.

He noted ADAS features including lane keeping and automatic emergency braking help cover for the driver, adding: “Long term, if we want to reduce accidents we need to address the data that says the driver is a significant part of the problem and the most likely solution would be to provide a substitute for the driver or a system that would take corrective action when the driver was not paying attention.”

Streelman also said if AVs are meaningfully able to reduce the US daily average of about 100 auto fatalities, self-driving cars may gain additional traction and grab fewer negative headlines. “What if the AVs were twice as safe as a human driver and the rate was reduced to 50 people in fatal accidents per day, would that be acceptable to the general public? We need to take into account how much attention an AV accident gets when a fatal accident occurs.”

The PAVE survey also found more than half of respondents would trust AVs more if they had to receive government approval like a human driver’s license. Until when or if regulators create a framework for AVs, Andy Whydell, vice-president of systems planning at ZF, explained: “The question is one of providing confidence to the general public that AVs can be trusted, and in the absence of regulatory definitions, each AV manufacturer will need to make their own decisions. This may lead to more aggressive manufacturers promoting their own metrics to try and differentiate themselves from competitors and delay more conservative and traditional manufacturers launching their own AVs.”

Some automakers are proactively reaching out to regulatory bodies in a bid to make sure both parties are on the same page and that their vehicles will ultimately gain approval. Whydell points to General Motors and Nuro each working with NHTSA as prime examples, with Nuro already gaining approval for its autonomous delivery vehicles.

The ZF executive also notes that UNECE regulations have delayed the introduction of Level 3 ADAS in some countries and he predicted Level 3 driving systems approval is likely still a few years away in Europe. As for testing AV safety, Whydell noted: “Simulation can fulfill several vital roles in the validation of AV testing, because it enables the use of consistent and repeatable test conditions to evaluate the performance and capabilities of AV systems in complete safety, especially in ‘edge cases’ or emergency scenarios that occur with low frequency on public roads and may be risky to reproduce.”

Clamann agrees that simulation works well and gives greater control over test conditions but adds there’s nothing like the real thing: “If you want the AV to go through the same intersection overnight seeing different vehicles, numbers of pedestrians, you can do that in simulation. Yet, only in the real world can you see the AV finds the sewer overflowing and can’t cross the water. You need both.”