SWIR Keeping a Closer Eye on Driver Monitoring Systems
Driver Monitoring Systems (DMS) are in-vehicle safety systems that use cameras to monitor the alertness and overall vigilance of the car’s driver while on the road. These systems react when signs of conditions like drowsiness or distraction are detected, triggering alerts to the driver, and initiating interventions to control the vehicle. In semi-autonomous vehicles, DMS have the potential to minimize the likelihood of an accident from distraction or fatigue and prepare the driver to take back control of the vehicle if the situation dictates. Unfortunately, these systems, which rely on a visible or NIR camera to monitor the driver, fail to operate in many common driving scenarios.
Driver distraction is a significant and difficult safety problem to address as it increases the chances of road accidents dramatically. In 2018, nearly 400,000 people in the U.S. were injured in crashes involving distracted drivers, according to an NHTSA report. Current DMS systems, that are intended to solve this challenge, suffer from significant inconsistencies that make their output unreliable for those who depend on the systems’ accuracy.
Where Do Existing DMS Fall Short?
DMS do not function properly under a wide range of scenarios, such as when the sun is shining into a car and the ambient light blinds the camera or when a street is covered with trees or buildings and creates shadow patterns that confuse the DMS algorithms. Another example is when reflections on glasses from different objects and vehicle headlights make it difficult for the gaze detection algorithms to differentiate between pupils and other reflections.
DMS are especially critical at night when there is a greater chance that the driver will become drowsy or unfocused on the road. To help the DMS camera function in low light conditions like nighttime, system architects need to implement a NIR illuminator which, for most people, will be invisible. However, these kinds of systems encounter situations that make them ineffective such as saturation from outside illumination and others. This leads to an inaccurate assessment of driver alertness by the monitoring algorithms.
But the shortcomings of existing DMS are not indicative of its legitimacy—it is here to stay and has an important role in the future of safe and reliable automated driving. Therefore, the technology needs to be perfected.
A SWIR-Based Driver Monitoring System
Short-Wave Infrared (SWIR) cameras have the potential to significantly improve DMS and provide more insightful image data:
Seeing in All Driving Scenarios – SWIR can be leveraged to solve the low visibility challenge for the car interior with full robustness to ambient illumination. It delivers enhanced vision even under the most challenging scenarios, with significant efficacy at night, when most needed. This function is enabled by operation in the solar-blind region, which prevents glare and shade patterns, combined with the eye-safe powerful illumination (SWIR can be used with 3-5 orders of magnitude more power and still be eye-safe, even in pulsed mode). This allows for an increased Signal-to-Noise Ratio (SNR) and therefore more precise and reliable real-time monitoring.
Remote Material Sensing – Another valuable benefit of SWIR to DMS is its unique ability to remotely sense materials. The SWIR spectrum provides a unique signature of target materials which results in the ability to recognize materials, rather than colors. This capability enables another layer of data to be used to monitor the driver’s health and provide life-saving advanced insights.
Using Existing Algorithms – Additionally, a SWIR camera’s output is similar to that of a standard visible camera, which enables the reuse of existing deep learning DMS algorithms without the need to develop, train, and annotate vast amounts of data. Thus, it allows for seamless integration, significantly shortening the time to market.
The TriEye Vision
TriEye’s world-first CMOS-based SWIR camera is revolutionizing machine vision capabilities by enabling crucial image data where the human eye or even a standard visible camera just cannot “see”. After leveraging our cost-effective, life-saving technology to enhance visibility on the road ahead, TriEye is offering this unique solution to enhance in-vehicle safety systems and dramatically reduce accidents on the road.
About the Author
Avi is the CEO and Co-founder of TriEye. He is an experienced multidisciplinary Physicist and Electrical-Engineer. Avi has a BSc in Computer Science and Electrical-Engineering and an MSc in Applied Physics specializing in optics and lasers.