On December 21, 2023, the Insurance Institute for Highway Safety announced the results of two new studies involving front crash prevention systems and how they detect vehicles that aren't other cars.
One study analyzed over 160,000 police-recorded crash reports from 18 states that met certain criteria, while the other study concentrated on how current New Car Assessment Programs (NCAPs) evaluate forward collision warning (FCW) and automatic emergency braking (AEB) features on new cars.
The results of both studies pointed to a need for significant improvements in how front crash prevention systems detect both trucks and motorcycles. Here are the most interesting things we learned from these studies.
Things Learned From The Multi-State Police-Reported Crash Analysis Survey
This study's more formal title is Are front crash prevention systems less effective at preventing rear-end crashes where trucks and motorcycles are struck?, and it was authored by Jessica B. Cicchino and David G. Kidd.
The authors analyzed data from over 160,000 two-vehicle rear-end crashes involving a passenger vehicle as the vehicle that rear-ended another vehicle. For simplicity's sake, we'll call this vehicle the Striker. In this study, Strikers could either have front crash prevention systems (such as FCW and/or AEB systems), or they could not.
It also looked at struck vehicle types in the crashes it analyzed, dividing them into three categories: Passenger vehicle, medium/heavy truck, and motorcycle.
In this analysis, the researchers used VIN decoder information to determine the correct category for the vehicles involved in these crashes when possible. This was helpful in correctly categorizing large pickup trucks, which were sometimes misclassified as "trucks" on police reports. (For categorization purposes, the researchers defined a medium/heavy truck as one having a gross vehicle weight rating over 10,000 pounds.)
Importantly, Cicchino and Kidd noted that front crash prevention systems were associated with a 53 percent reduction in rear-end crashes that involved the Strikers hitting other passenger vehicles. However, that percentage reduced significantly when it came to avoiding rear-end crashes with medium/heavy trucks and motorcycles.
The analysis, which included data from 18 states, found only a 41 percent reduction in rear-end crash rates where the struck vehicle was a motorcycle. With medium/heavy trucks, the number was worse; only a 38 percent reduction was observed.
By their estimates, the researchers wrote that around 5,500 additional crashes involving medium/heavy trucks and 500 crashes involving motorcycles could potentially be avoided if front crash protection systems were improved so that they recognized medium/heavy trucks and motorcycles as well as they currently recognize other passenger vehicles.
Furthermore, they observed that "nearly half of motorcycle crashes are two-vehicle crashes where the other vehicle was a passenger vehicle."
Beyond that, though, came another chilling observation that didn't involve rear-end crashes of motorcycles at all.
Here, the researchers wrote "For example, Teoh (2023) reported that over a quarter of two-vehicle motorcycle crashes involved the other vehicle turning left in front of the motorcycle, which could be addressed by left-turn assist systems that detect motorcycles." In other words, left-turn assist systems doing the heavy lifting of seeing motorcycles when drivers, for whatever reason, simply do not.
Things Learned From The Surrogate Targets Survey
DRI Soft Motorcycle 360 Surrogate Vehicle Target
4active Systems 4activeMC Surrogate Vehicle Target
This study's formal title is The effectiveness of forward collision warning systems in detecting real-world passenger and nonpassenger vehicles relative to a surrogate vehicle target. It was written by David G. Kidd, Benoit Anctil, and Dominique Charlebois, and was the result of a cooperative effort between the Insurance Institute for Highway Safety and Transport Canada.
For this study, rather than analyzing data from real-world police reports of crashes, the analysis instead concentrated on front collision warning performance trials of the type used in NCAP testing. These trials typically involve the vehicles to be tested crashing into either other vehicles or what's known in the testing industry as 'surrogate vehicle targets.'
Why use surrogate vehicle targets? In many cases, they're constructed to show up on the radar, lidar, camera, and infrared systems typically used in Advanced Driver Assistance Systems (ADAS), a category which includes front crash prevention systems. At the same time, they're also soft and therefore less likely to damage the vehicles that are undergoing testing if they don't stop in time.
In this study, two purpose-built surrogate vehicle target motorcycles were included: the DRI Soft Motorcycle 360 and the 4active Systems 4active MC. Both of those vehicles are currently used in real-world testing by various NCAP programs around the world.
A 2006 Honda VFR800 was also included among the vehicles to hopefully be avoided by the passenger vehicle front crash prevention warning systems that underwent testing. Additionally, a Polaris Slingshot was among the other non-motorcycle vehicles included as a target to be avoided by the passenger vehicle FCW systems being tested.
In describing a 2016 study performed by John Lenkeit and T. Smith, the researchers observed that, "[The authors] evaluated FCW systems in eight vehicles when each was traveling 72 km/h (45 mph) and encountering a stationary motorcycle (2006 Honda VFR 800-cc sport touring) or midsize passenger car (2000 Honda Accord). Five of the eight FCW systems either did not consistently warn for the stationary motorcycle or provided a warning later than 2.1 seconds before collision in more than 2 trials; 2 of the vehicles never provided a warning. In contrast, every FCW system detected the stationary midsize car and provided a warning 2.1 seconds prior to collision or earlier."
That wasn't the case when the testing vehicles approached slower-moving motorcycles rather than motorcycles that were completely stationary.
"Lenkeit and Smith also tested FCW performance in four of the eight vehicles when the vehicle was approaching the same motorcycle or passenger car moving slower (32 km/h [20 mph]) than the test vehicle (72 km/h [45 mph]). In contrast to the stationary vehicle results, all four vehicles detected the slower moving motorcycle and provided a warning 2.1 seconds before collision or earlier. These findings suggest that the sensors and algorithms supporting FCW struggle to identify stationary motorcycles but not stationary passenger cars."
Keep in mind, the quoted study was from 2016. It is now the end of 2023, and it's also very nearly 2024. Presumably, ADAS technologies have advanced at least somewhat since the time that this initial study was conducted. However, it's still safe to say that these observations identify areas where serious improvement is needed.
That test from 2016 relied on a real motorcycle that was 10 years old at the time of testing. More recent testing involving surrogate vehicle target motorcycles yielded the following note in this study.
"Consistent with past research, FCW systems struggled to detect and provide timely warnings when the vehicles approached a stationary motorcycle in the center of the lane. On average, the FCW systems presented a warning 0.18 and 0.15 seconds later for the 4activeMC and DRI soft motorcycle 360 than the DRI GVT, respectively. Motorcycles are smaller than passenger cars and have a smaller radar cross section that produces less radar signal return, which may delay or negate detection by radar sensors. Camera sensors also may struggle to detect motorcycles because their smaller size contributes to lower spatial resolution in the camera image, especially at further distances. Higher resolution cameras are important for reliably detecting smaller objects."
Two more interesting observations we noted in this study regarded the Slingshot, about which the researchers noted that,
"The Polaris Slingshot is an autocycle with a single rear wheel like a motorcycle and a cockpit and two front wheels like a passenger car. The Slingshot’s mixture of motorcycle and passenger car features appeared to confound the FCW systems that were tested. A warning was provided in only a little more than one-third of the trials. But when the Slingshot was detected, the timing of the FCW was statistically equivalent to the DRI GVT. Hence, the FCW systems appeared to treat the Polaris Slingshot as a passenger car when it was detected."
But overall, as was the case in the multi-state real-world crash survey, the researchers came to the conclusion that,
"Responses from FCW systems in modern vehicles to stationary nonpassenger vehicles like motorcycles and large trucks were not statistically equivalent to how the systems responded to a stationary passenger-car surrogate target. The findings emphasize a need for developers to train object detection algorithms underpinning FCW and AEB systems on a broader set of potential vehicle target classes and not only the most common vehicle types or those that are evaluated in vehicle testing programs."
While it's not realistic to expect testing bodies to take every single size and shape of motorcycle (or other non-passenger vehicle) into consideration, some variety of motorcycle sizes and shapes would probably be helpful. There's a world of difference between, say, a Honda Grom and an Indian Chieftain.
Meanwhile, the Honda VFR800 and the two surrogate vehicle target testing 'motorcycles' used in the second survey all have similar size and shape characteristics. The DRI Soft Motorcycle 360 has a wheelbase of 1,460mm, or 57.4 inches. That is, in fact, the same wheelbase as the VFR800, or to give you a modern wheelbase equivalent, a 2024 Suzuki GSX-S1000GT+.
While that's certainly not an uncommon general size and shape for a motorcycle to have, it's far from the only one. As any rider can tell you, there's far more than one way to enjoy motorcycling, and many of them do involve sharing the road with other vehicles. The wheelbase on the 4activesystems model is slightly smaller, at 1,420mm, or about 55.9 inches, but there still needs to be greater variety in testing and detecting.
Advances in front collision warning systems on passenger vehicles are encouraging, but as the IIHS illustrates here, improvements are still necessary. Truck drivers and riders alike all want to get to wherever we're going safely, and if improved electronic driver aids can help make that happen, we're all for it.