By Paul Morris
Cameras are being used in more vehicles every day, and while most current applications look "outward" (like rear-view camera or lane-departure-warning) newer implementations will face "inward" – capturing views of the driver, passengers and cargo. Visteon has been developing driver monitoring applications and recently introduced the "HMEye" demonstration concept. The name HMEye is a play on the term Human Machine Interaction (HMI) combined with Eye-Tracking. The concept uses eye-tracking cameras and software that allow the driver to interact with some vehicle controls.
Some in-vehicle camera products available today provide features like drowsy and distracted driver monitoring. These cameras observe the driver’s eyelid opening, blink rate or head position. Most of these vehicles use a single camera and some type of infra-red (IR) illumination to work during the day or at night. For example, Lexus offers a driver monitoring system as part of its Pre-Collision Safety Package and other companies offer aftermarket solutions for large truck fleets.
For a system to monitor the driver, it needs to be positioned in front of the driver’s face, which is why wearable trackers can be advantageous (see image below of a wearable tracker). However, drivers usually won’t be wearing a tracker, so a camera with IR flash system needs to be mounted in the direction the driver intends to use it. Typically, it will be affixed in the forward-looking direction, but could also be positioned to the side, depending on the range of eye/head movement the system is designed to track. Low-cost integrated trackers will be limited in vehicles, because they usually are only configured for a single display on a desktop. The main challenge for the vehicle interior designer will be to package the cameras and flash units in a way that is optically functional, but also aesthetically pleasing.
Once enough camera and flash coverage is provided, a higher-accuracy gaze-tracking function can be added to enable features like gaze-controlled HMI. This feature is demonstrated on Visteon’s HMEye demonstration property. These features would be in addition to the base drowsy and distracted driver features mentioned above.
There are many interesting scenarios when capabilities like computer vision with artificial intelligence (AI) are added. Think about the vehicle as an “attentive mobility assistant” that always focuses on its customers: the driver, passengers and cargo in the vehicle. In the past, the vehicle only responded to the sense of touch (a button press, etc.). A few years ago it evolved into hearing (via microphones) and can now understand vocal commands. Vehicles are integrating sight (via cameras), which initially will be poor, but will improve over time as they are developed to recognize new objects and behaviors. A vehicle will see externally (e.g. pedestrian detection) and internally (e.g. driver monitoring) and will begin to learn patterns from good drivers and share them with drivers who need more help. The system will aid the transition between driver-control and autonomous-control and have context and situational awareness that allows it to adapt its behaviors appropriately. Many new opportunities will emerge as vision technologies continue to develop.