Eyes on the Driver: How AI Can Keep You Safe Behind the WheelAugust 4, 2017
Tailgaters. Red light runners. Hard braking. Whether in a Mini or a Mack truck, every driver faces nerve-wracking moments.
With a small, AI-powered device mounted on the windshield, unsafe driving events like these can be detected as they happen and their causes analyzed.
Avneesh Agrawal, CEO and founder of Netradyne, explained how in a presentation at the GPU Technology Conference in May. His San Diego-based startup has coupled deep learning and edge computing with a device dubbed “Driveri™.”
Mounted behind the rear-view mirror in cars, vans, trucks or big rigs, Driveri acts like a computerized command center. It uses real-time analysis from data taken from cameras, vision-based deep learning and inertial sensors mounted on a vehicle to put “eyes” not just on the road, but also on the driver, the car and everything around it.
Its goal: recognize great driving and improve commercial fleet and driver safety.
Diving Into Driver Analytics
Driveri can help transform safety in all aspects for motorists, most especially commercial drivers, whose own livelihood and safety of those around them depends on their care and attention. Its applications can recognize good driving — and point to areas where best practices need to be reinforced.
“We’re all looking to self-driving cars, but the reality is that by 2030, less than 10 percent of vehicles on the road will be autonomous, so for the next 10 to 15 years we have to invest in drivers,” Agrawal said. “The return on that investment shows up immediately, in creating safer driving.”
Driveri uses the NVIDIA Jetson TX1, a powerful credit-card sized supercomputer embedded on a module. This is coupled with a vision-based, 360-degree camera video system, which captures everything surrounding the vehicle, as well as nearby traffic signs and signals, other vehicles and pedestrians.
“It essentially sees everything the driver sees and experiences,” Agrawal said. The data is analyzed at the device and data is uploaded to the cloud. A software-as-a-service customer portal provides information to the fleet within minutes of an event.
Risk and Reward
Leveraging the processing power of the Jetson TX1, Driveri can capture every aspect of the driving day, creating a score on the entire day’s driving versus only looking at a small sub-segment.
The data rolls up into Netradyne GreenZone™ score, a real-time and comprehensive record of driving performance that fleet managers can use to create incentives for safer driving. Insurance companies can also use the data to set premiums customized for individual drivers and different driving environments.
The analysis can also help establish context for different situations that simple alert systems can’t show. For example, capturing the full sequence of events leading up to an incident can reveal if hard braking was caused by a driver who was tailgating or by another motorist swerving in front of the driver’s car.
“Commercial fleets experience large driver turnover, which can cost about $8,000 per driver a year,” Agrawal said. “We want better risk assessments, where you can identify at-risk driving and create an incentive structure based on good driving.”
Netradyne has already captured and analyzed over 1 million miles of driving data with its customer community. It’s now looking beyond commercial use cases to consumer ones.