Researchers at Princeton University have developed new radar technology that should make the roads safer. The key is improved AI, that can “see around corners.”
Radar-equipped vehicles are the way of the future, with self-driving car technology already filling the roads, and the motorcycle industry developing similar designs. Most autonomous driving systems use a combination of inputs, including cameras, accelerometers, lidar and radar to keep a vehicle from crashing. Radar, especially, is the key of new active cruise control systems. BMW, Harley-Davidson, Ducati and Kawasaki are all working on these designs.
Current onboard radar systems have limitations, though. They scan in a straight line; the radar doesn’t detect something that’s not in its direct line-of-sight. A team of researchers from Princeton realized they could improve motoring safety if they could somehow “see” around corners with the vehicle’s sensors, so they went to work developing a solution.
Safety through software
First, they worked on lidar technology, using high-powered lasers to detect objects hidden around corners. However, the researchers say this design was too expensive, and only worked at shorter ranges. Instead, they decided to focus on using imaging radar.
Imaging radar has several advantages over lidar. It’s cheaper, and has decades of use in civilian and military tracking applications. The problem, according to Princeton, is that “radar’s spatial resolution — used for picturing objects around corners such as cars and bikes — is relatively low.” So, the wonks went to work. Instead of drastically redesigning the radar hardware, they developed software algorithms that interpreted the radar data more efficiently. Now, instead of rejecting a jumble of junk information, the autonomous driving systems can use that information. The software processes information from radar signals “bounced” around corners, and uses that data to tell motorists what’s ahead.
While the technology seems to have been developed for cars, this is good news for motorcyclists for a couple of reasons. First, if auto AI “sees” motorcyclists and can avoid crashing into them, that will improve our safety. Second, this tech may be car-only for now, but sooner or later, it will trickle down to the world of motorcycles.
And, more good news. Because this is mostly a software re-design, not hardware, Princeton’s researchers say it could come to market quickly, possibly in the next generation of autonomous vehicles. Maybe even on the next-gen motorcycles, with their adapative cruise control? Let’s hope.