The promise of self-driving cars shuttling passengers across roadways is seemingly creeping closer to reality. But, self-driving cars perform at their best in dry conditions without precipitation. Ask a current self-driving car to navigate a Midwestern snowstorm and it will not go well. A company called WaveSense may have the solution.
The company has developed complementary technology to today’s GPS and lidar to penetrate and scan roadways even during a snowstorm or other inclement weather. The company calls it “localizing ground penetrating radar,” and it works by scanning what’s under the road to keep a vehicle in its lane if road markings are not clearly identifiable.
When a self-driving car operates, lidar and cameras look for road markings to keep the car on track, while GPS can place a vehicle in roughly a 10-foot vicinity. As soon as these markings go away, however, the car goes blind and can stray from its lane. LGPR sends very-high-frequency (VHF) electromagnetic pulses 10 feet into the ground to create maps of subterranean road compositions. Those maps are cross-checked via the cloud against WaveSense’s continually updated subsurface maps. Thus, the self-driving car can identify lanes and keep lane meandering to a minimum.
WaveSense LGPR self-driving technology in the snow
The company says the scans can penetrate snow-covered roads, fog, heavy rain, and dust—all things today’s self-driving cars can’t master. And the technology has proven quite successful so far. WaveSense says the technology has demonstrated lane accuracy within 1.6 inches on snow-covered roads at 60 mph. Comparing a self-driving car’s performance at night in the snow versus a baseline data set during the day on clear roads showed a nearly identical track. Impressive stuff.
We imagine it’s only a matter of time before WaveSense’s technology makes its way into more self-driving car prototypes. If automakers and companies ever want to make safe autonomous vehicles a reality across the world, they’ll need to be smart enough to handle Mother Nature.
