Israeli far infrared (FIR) technology startup AdaSky’s thermal, high-resolution sensor for autonomous vehicles is the first to be integrated into smart headlights.
FIR sensor Viper has been integrated into Magneti Marelli’s Smart Corner Modular Platform, leveraging the energy efficient, high-performance compute of the NVIDIA DRIVE platform. Until this partnership, FIR cameras were mainly integrated into the vehicle’s grill. Integration into the headlamp means the sensor is closer to other sensors, which enables better fusion. Sensor protection and cleaning is easier, too.
The new computer-vision capabilities of Viper include multiclass object detection and classification, which enables a vehicle to simultaneously detect and classify pedestrians, vehicles, trucks, bicycles and motorcycles. Viper can also detect and segment animals and other objects in the vehicle’s surrounding environment.
At CES, AdaSky gave in-vehicle driving demonstrations to showcase the image quality of its FIR thermal sensor, which delivers real-time FIR-based perception to autonomous vehicles. Viper passively collects the FIR signal that radiates from objects and other materials and converts it to a VGA video. It then applies AdaSky’s proprietary deep-learning computer-vision algorithms to provide accurate object detection, classification and scene analysis.
“There is a growing need for our FIR solution. Even two years ago many OEMs were reluctant to adopt this technology,” says Yakov Shaharabani, CEO, AdaSky. “But now almost any OEM who is in front of the automotive space is seriously evaluating FIR, and we have plenty of projects in place with OEMs.”