Many automakers have features that allow a vehicle camera to monitor the driver and set off alarms when they fall asleep. However, researchers from the Fraunhofer Institute have created a smarter system inside the car camera system that can identify exactly what the driver is doing. This could improve safety features like semi-autonomous driving.
There is no publicly available self-driving vehicle that can handle all road conditions. Drivers must take over the wheel again when there are unexpected interruptions, such as construction delays or accidents. This handoff can be difficult. The ideal scenario would be for the semi-autonomous vehicle's driver to always pay attention to the road ahead and be ready to take control at any moment. Humans will still be human, so there is always the possibility that an autonomous vehicle will attempt to take control of the vehicle while the driver is distracted by a phone call or drinking coffee. This could lead to dangerous situations.
These situations were what inspired researchers at the Fraunhofer Institute of Optronics and System Technologies and Image Exploitation (FIO) to expand the capabilities of in-car camera systems. The new system uses AI-powered photo recognition to create a digital skeleton that looks almost like a stick figure. The digital skeleton is a simple representation of the current driver's pose. However, it provides enough detail for the system to understand what the driver is doing. Additional object recognition allows the system to keep track of the location of objects such as smartphones and coffee cups.
The vehicle can determine whether a driver is paying attention on the road or distracted by other activities such as texting, eating or interacting with others in the vehicle. Semi-autonomous driving systems can monitor the driver's activities and determine how long it takes them to focus again on driving.
This technology is not only making semi-autonomous driving safer. Although many vehicles can park themselves, drivers cannot simply speak to the vehicle and expect it to find the spot. Although voice commands are not always clear and precise, a camera system can detect where drivers are looking and possibly even point their fingers. This allows for other cues to be considered, such as knowing the exact parking spot that a driver is referring to. Smarter camera systems may also be able to assist with seat belt detection systems. These systems, at the moment only determine if the buckle has been secured and not whether a driver or passenger are wearing it properly.