Guest commentary: Explore sensor tech to improve nighttime braking, save lives

In recent years, amid heightened concerns surrounding vehicle safety, manufacturers have turned to sophisticated technologies, such as automatic emergency braking, to reduce or mitigate the severity of crashes. And while automatic emergency braking, or AEB, appears to be making a positive impact, a study from the Insurance Institute for Highway Safety reveals that these systems are still far from perfect — particularly when it comes to detecting pedestrians at night.

In the nighttime test for pedestrian AEB, or P-AEB, more than half of the first 23 midsize cars, midsize SUVs and small pickups tested earned a basic score or no credit, while only four earned a superior rating.

Jessica Cicchino, vice president of research at IIHS, described it as "the first real-world study of pedestrian AEB to cover a broad range of manufacturers." The organization found that while vehicles equipped with P-AEB systems reduced the odds of a crash by more than 30 percent in well-lit conditions, they offered no identifiable reduction of risk to pedestrians at nighttime in areas without streetlights.

As IIHS was careful to point out, this is a significant problem when you consider that around 75 percent of fatal pedestrian crashes happen at night. And perhaps needless to say, it is a problem that manufacturers and engineers in the autonomous mobility industry can't afford to leave unaddressed.

So where are these systems coming up short? To answer this question, we first need to take a closer look at the different technologies being used in existing P-AEB systems.

P-AEB technologies found on vehicles today, and therefore included in the IIHS study, use either cameras, radars or a combination of both. As cameras collect ambient light radiation to detect and track objects, their performance significantly degrades in dark lighting conditions. Radars, in contrast, send a radio frequency signal and collect the return signal after it is reflected off an object. Due to the operating frequency of automotive radar, 24 gigahertz or 77 GHz, it is not impacted by lighting conditions nor adverse weather conditions. However, most radars currently found in P-AEB systems lack the performance needed to accurately detect and track a pedestrian at night.

There are two primary candidates for improving nighttime P-AEB systems: lidar and digital imaging radar. While both technologies attempt to solve the same problem, the differences between the two are substantial and worth exploring.

Lidar is a laser-powered sensing technology that uses light to scan its surroundings and construct a map of the environment. Lidar has roots in space travel and meteorology but has more recently been utilized in the autonomous mobility space, most famously by its inclusion in Google's first self-driving car.

Unfortunately, lidar does have some significant limitations that engineers and manufacturers will find difficult to overlook. For one thing, lidar is not a proven automotive technology, which means it will require a great deal of continued development and operational reliability refinement to be a viable solution in this context.

Manufacturers also are likely to be turned off by the high cost of lidar sensors compared to radars and cameras. But perhaps the biggest drawback of lidar is the fact that, despite performing well in the dark, it is not robust to the broader volatility of an environment and can fail to produce accurate images in harsh weather conditions.

A second option is to replace current analog radar sensors with digital imaging radar systems. Digital imaging radar can be thought of as the logical evolution, or next generation, of its analog counterpart. Digital imaging radar, contained in a single radar-on-chip, can detect and track thousands of objects in real time and produce high-resolution output. Unlike lidar, it has already been deployed in autonomous ground vehicles.

Beyond enabling P-AEB to function in the dark, digital imaging radar also has the potential to mitigate other safety concerns uncovered in the IIHS study. For example, while AEB systems reduced crash risks by more than 30 percent on roads with speed limits of 35 mph or less, they failed to reduce risk at all on roads with limits of 50 mph or higher. Because digital imaging radar can detect both pedestrians and vehicles from up to 300 meters away, AEB systems would have more than enough time to trigger the brake function and avoid a collision.

While the superiority of digital imaging radar compared to virtually any other AEB system or alternative seems clear, the exploration of all possible solutions is imperative to the collective goal of safer roadways for pedestrians and drivers alike.