Scientists at the University of Michigan have introduced a cost-effective hardware approach that leverages current technologies, incorporating cameras equipped with facial recognition and analysis features.
This method integrates conventional Advanced Driver Assistance Systems (ADAS) cameras that are used to monitor various aspects of driver behavior and alertness, with either infrared LiDAR or structured light 3D cameras, which capture detailed 3D images and measurements.
“You already see these 3D camera technologies in products like smartphones, tablets, and mixed reality devices,” says Mohammed Islam, U-M professor of electrical engineering and computer science who leads the project.
The National Highway Traffic Safety Administration (NHTSA) has outlined the Federal Motor Vehicle Safety Standard (FMVSS) that will require new passenger vehicles to be equipped with advanced drunk and impaired driving prevention technology, such as the low-cost hardware solution proposed by the University of Michigan.
This effort is part of the enforcement of the Bipartisan Infrastructure Law that targets the reduction and eventual elimination of fatalities due to alcohol-impaired driving. The report indicates nearly 30 percent of deaths are alcohol-related.
The development at the University of Michigan is a direct response to this new federal mandate. The requirement that all new passenger vehicles include these protective measures could be implemented as early as 2026.
“These 3D cameras are small, inexpensive cameras that can easily be mounted on the rearview mirror, the steering column or other places in the driver’s cockpit.
The approach depends on integrating LiDAR or structured light 3D cameras into the existing vehicle safety systems. The estimated cost for this hardware solution is projected to be between $5 and $10 per vehicle.
Researchers claim that the significant cost savings of the facial recognition system will offer a competitive advantage over traditional in-vehicle breathalyzers, which can cost around $200 per vehicle. This suggests that the proposed solution will be a more economically viable option for mass adoption.
Additionally, the hardware will be enhanced by artificial intelligence analysis, which will interpret the complex biometric data captured by the cameras. The data would focus on identifying specific physiological and behavioral indicators of impairment.
These can include increased blood flow to the face, determining the heart rate, monitoring eye movement and behavior, observing changes in head position and body posture, and observing respiratory rate. By applying machine learning algorithms, the system can analyze deviation from a baseline measurement.
“In many new vehicles, Advanced Driver Assistance Systems (ADAS) cameras are already onboard to track driver alertness. They’ve already been matured and are cost-effective.”
To further develop this technology, the research team is working closely with Tier 1 automotive suppliers like DENSO and Tier 2 suppliers that are experts in the production of camera technologies.
A key benefit of this system is its non-invasive and passive nature, which operates without requiring active participation from the driver. This feature is likely to decrease user resistance toward adopting the technology. It also makes it more difficult to spoof by having someone other than the driver evaluated, and performing assessment continuously.
The researchers also emphasize that 3D cameras with infrared capabilities can address two significant challenges traditional camera systems face in monitoring driver alertness: difficult lighting conditions and precision in motion tracking.
Because 3D cameras can focus on infrared light emitted from them, they are unaffected by changes in ambient lighting. This eliminates the challenges associated with low-lighting conditions for in-vehicle safety systems.
Moreover, the capability of 3D cameras to gather depth information and accurately track the driver’s 3D motion allows for a clear differentiation between actual changes in driving conditions.
A study led by Mohammed Islam in 2022 introduced a contactless vital sign monitoring (CVSM) system for measuring in-cabin heart rate and respiration rate using a near-infrared indirect Time-of-Flight (ToF) camera.
The goal was to address the challenges of varying ambient illumination and interference from excessive motion. Islam says, “Our results show that our CVSM system is ambient light agnostic, and the success rates of HR measurements on the highway are 82% and 71.9% for the passenger and driver, respectively.”
Researchers believe that introducing 3D infrared cameras for driver monitoring systems can address the challenges with existing integrations and offer a cost-effective solution.
Article: Scientists use biometrics, behavior analysis for drunk driver detection