As motorcycle riders, we’re now sharing the road with more and more highly automated vehicles.
Tesla’s vaunted “Autopilot” function makes it the best-known HAV, but quite a few new cars combine a variety of forward collision warning (FCW), automatic emergency braking (AEB), and lane-assist functions. If you believe Tesla, its advanced driver assistance systems (ADAS) have already prevented thousands of accidents. From Tesla CEO Elon Musk’s perspective, the glass is HAV full. But if you’re a motorcyclist, you probably see the glass as HAV empty.
On July 27, 2016, a Norwegian motorcyclist was out riding, minding her own business, on Norway’s busy A18 freeway near the city of Drammen. She was rear-ended by a guy “driving” a Tesla Model S. Although she survived the crash, she was seriously injured. I use the sneer quotes on the word “driving” because while the driver was behind the wheel, his Tesla was in Autopilot mode. The car’s sensors and software failed to identify the woman on her motorcycle as something it should slow down for or avoid.
Norway, eh? It turns out those socialist Scandinavians have a national policy that aggressively encourages electric vehicles, and Norway’s an important export market for Tesla. So I suppose it’s not that surprising that the first high-profile collision between a Tesla in Autopilot mode and a motorcyclist happened there.
The Norwegian equivalent of the AMA is the Norsk Motorcykkel Union (NMCU). They were understandably alarmed to learn that a car that’s common on Norwegian roads — and which is marketed as having an Autopilot feature — couldn’t seem to recognize a slower or stopped motorcycle as another vehicle and avoid smashing into it.
The NMCU’s General Secretary, Morten Hansen, wrote a letter of concern to Norway’s Minister of Transportation, with a copy to Elon Musk. “It is worrying if Tesla’s ‘autopilot’ is not able to detect a motorcycle, but it is even more worrying if the driver has been duped into believing that a driver assistance system is an ‘autopilot’ that allows you to cross your arms and no longer be the driver of your car,” Hansen wrote. “NMCU is certainly not against a future of autonomous vehicles, but we are a bit concerned that manufacturers and politicians are ‘over-selling’ present technology.”
Considering that a Norwegian motorcyclist was nearly killed, I think that letter was a model of self-restraint.
The NMCU’s action also inspired the Federation of European Motorcyclists’ Associations and two Dutch motorcyclists’ associations to write a more stongly worded letter to the Rijksdienst Wegverkeer (the Netherlands Vehicle Authority, which is the Dutch equivalent of the federal Department of Transportation, National Highway Traffic Safety Administration and your state Department of Motor Vehicles all rolled into one).
Tesla’s European PR problem got a lot worse when the influential German weekly Der Spiegel got its hands on a research report prepared by that country’s Federal Highway Research Institute (BASt) on the Tesla Model S Autopilot. Basically, Germany’s version of NHTSA concluded that the Autopilot function was nowhere near up to the task of driving on the Autobahn, where cruising speeds of 120 mph are common.
BASt went as far as to recommend to the German transport minister that the Tesla’s “type certification” be removed. The government didn’t go that far, but German Tesla owners all got a letter warning them to pay attention in Autopilot mode. The Norwegian incident, and several other high-profile Autopilot crashes (most of which did not involve motorcycles) caused Tesla to temporarily dial back Autopilot functionality; for example, the company reduced the top speed in that mode.
Meanwhile, a motorcyclist in California…
John Lenkeit is an engineer who works for Dynamic Research – a consulting firm specializing in “accidentology,” among other things – with offices in Torrance, California, and a proving ground on the site of an abandoned airport near Bakersfield. One of the company’s ongoing gigs is conducting NHTSA-mandated tests on new cars destined for the U.S. market. In particular, they test Forward Collision Warning and Automatic Emergency Braking systems.
Lenkeit’s the guy in charge of that kind of testing, some of which involves driving data-logged test cars towards target cars that are either stopped or driving slowly ahead. The tests record if and when the cars’ FCW alarms are triggered. Then, if so equipped, the tests watch for the application of the Automatic Emergency Brakes. The tests aren’t destructive; drivers swerve out of the way in situations where the automated systems aren’t sufficient. Most new cars that include such systems devote quite a few pages to warnings about them in the owner’s manual. Those warnings often include a caveat that the systems may not detect motorcycles.
John beats SoCal traffic into work by commuting on a Honda VFR800, so he’s personally aware of the extent of distracted driving on American freeways, as well as professionally aware that FCW and AEB systems prevent a lot of collisions. It was probably inevitable that he’d eventually wonder, “Would such systems protect me from being rear-ended in traffic?”
Around the time that Norwegian was getting rear ended by the Tesla, John started taking his VFR up to the company’s Bakersfield strip. On days when there was still time available after the conclusion of NHTSA tests, and while the subject vehicles were all still datalogged, they simulated the NHTSA test using the VFR as the “principal other vehicle” (instead of the usual target, a Honda Accord.)
The results, as he wrote to me in an email, “might scare you.”
Basically, in testing over half a dozen different cars from various manufacturers, the FCW systems failed about 40 percent of the time. That compares to a typical failure rate of about 4 percent in simulations involving the Accord.
I should note that Lenkeit anonymized the results of his research before presenting it, so I do not know which brands of cars he tested. That said, the mix of hardware used in all manufacturers’ detection systems is conceptually similar. There’s every reason to conclude that Lenkeit’s results are typical of all cars that are so equipped.
Lenkeit and a co-author, Terrence Smith, published their findings in a research paper titled, “Preliminary Study of the Response of Forward Collision Warning Systems to Motorcycles.” They presented it last October in Cologne, at a conference sponsored by the German Institut für Zweiradsicherheit (‘Institute for two-wheeled safety’).
Although Lenkeit’s the first to admit the Dynamic Research study is preliminary, and that the data set was small, the implications of the study are large, especially for motorcyclists.
“Detection of motorcycles by other vehicle drivers is a constant challenge in the current riding and driving environment,” he warned. “As drivers become comfortable with, and rely more on ADAS [advanced driver assistance systems] technologies, they may become less attentive to the driving task and thus be less vigilant at detecting motorcycles on the roadway. If ADAS systems are unable to correctly identify motorcycles, a possible consequence of broad ADAS implementation may be an increase in car-motorcycle accidents even as car-car accidents decrease.”
In conclusion, he also noted that “for the safety benefits of ADAS systems to extend to motorcycles, such systems need to reliably detect motorcycles in potential crash scenarios. One way to encourage and verify this would be to include motorcycles or their representations in ADAS test procedures.”
Let me translate that from academese, for you. What John means is, Hey! Motorcycles are out here on the roads. If auto makers are going to release and market cars with any degree of self-driving capability, automakers and regulatory bodies should ensure that those cars won’t just run us over. Because right now, all they’re doing is lulling car drivers into a false sense of security where car-motorcycle collisions are concerned.
To be clear: Many manufacturers are currently testing HAVs on public roads. You can be sure they’ve privately logged many interactions between HAVs and motorcycles. But the organizations you’ve entrusted with your basic safety — for example the Department of Transportation and National Highway Transportation Safety Administration — aren’t testing current ADAS systems’ ability to detect motorcycles when they approve new cars. And as we move towards fully autonomous vehicles, there has so far been an alarming absence of input from the motorcycle world.
One of the last things that the DoT and NHTSA did under the Obama administration was to issue a Federal Automated Vehicles Policy, a document more than 100 pages long. The words “motorcycle” or “motorcyclist” are not found anywhere in the document. Even European agencies — in countries where there are more motorcycles on the road, where motorcyclists are more politically active, and where regulators are more pro-active — have largely left motorcycles out of HAV testing protocols.
What if they ban us to avoid hitting us?
Lenkeit speculated that many agencies test cars vs. cars, and then cars vs. pedestrians and bicyclists; then they assume that since a motorcycle’s somewhere between those two extremes, any system that will avoid hitting cars and pedestrians will also avoid hitting motorcycles. It’s not that simple, in part because one big issue with ADAS technology is avoiding false positives, which might cause a car to suddenly brake for no good reason.
Quite a few motorcyclists don’t just feel that the glass is HAV empty, they think the glass is about to be confiscated. Every time I read or write about autonomous vehicles, I get comments similar to this from Reddit poster “smokeybacon0149” (a UK contributor who rides a BMW S 1000 R): “I imagine filtering is really going to fuck them up and piss off the car's occupants when it brakes suddenly because of a bike zipping by. If so, that really won't do your campaign to legalise filtering any favours.”
Translating from British, that means it really won’t do our campaign to legalize lane-splitting any favors.
Another Reddit poster and Yamaha FZ-09 rider, “victheone,” wrote, “Having all cars be autonomous will be much safer for everyone, motorcyclists included. It'll be great until a motorcyclist causes an accident and the public outcry forces the government to ban motorcycles. People already think motorcycles are stupid and unsafe, imagine what they'll think once we are the only ones on the road actually controlling a vehicle. Humans don't like things that are ‘different.’ They'll move to ban riding.”
The reality is that many new cars and trucks will soon be self-driving. Too many companies have invested too much money in autonomous vehicles. And too many people – Baby Boomers for whom a car is a critical part of their identity – will refuse to give up their cars as they age. Companies will push self-driving cars into the market. Aging customers will snap them up. And the federal government’s in no mood to say, “Hold on a minute… Are these things safe?”
What if the glass is HAV full?
The technology’s clearly far from perfect, but there’s simply no stopping it — or even slowing it. If you’re a “glass-HAV-full” kind of rider, you’ll welcome our robot overlords. Autonomous tech may not be perfect but let’s face it: human drivers are terrible. Autonomous vehicles will never get distracted by an incoming text or the kids in the back seat and they will never be drunk or stoned. But when it comes to motorcycles, robot drivers face the same challenges human ones do: Motorcycles are smaller than most of the other vehicles on the road and we have a tendency to appear in unexpected places, at unexpected speeds.
BMW, is one of the few auto makers that also has a significant motorcycle business. Motorrad’s head of development, Karl Viktor Schaller, recently told Australia Financial Review that “With more and more autonomous cars on the road, people will be looking for new ways to pursue freedom,” adding that autonomous cars “would mean a dramatic enhancement in safety for the motorbike, and it would guarantee a wider user group.”
So, Schaller and BWM are in the glass-HAV-full camp. And for all the flaws Lenkeit discovered — to say nothing of the flaw that woman in Norway discovered — there are reasons to be optimistic.
In my next post, I’ll review efforts being made by organizations like the AMA and Motorcycle Industry Council to protect motorcyclists’ interests, and report on technology that promises real improvements to rider safety. So, check back in a few days for more of the glass-HAV-full perspective.