Automatic Driver Assistance Systems
December 15, 2014, IEDM, San Francisco—Hideo Inoue from Toyota talked about the potential future innovations in automatic driver assistance systems in automobiles. The advances in processing power and continued device shrinks will enable more integrated assist, safety, and automation in future cars. The key is the increased intelligence available.
One area is a safety item. Pre-collision systems will include avoidance capabilities in addition to the automatic speed reduction and braking functions now available. The addition of IR sensors will extend the driver’s view and will link to the brakes. The increasing number and function of the various sensor systems depend on more electronic.
Driver assist technologies are moving from mitigation technologies like seatbelt pre-tensioning to the emerging emergency avoidance and will eventually work to minimize human errors. The increase in sensors and processing will minimize the least reliable component in the car, the driver.
Intelligent driving technologies can help the driver by providing safe route detection and tracking. The addition of LIDAR and other sensors enables the integration of more capabilities that can help a subset of the driving population, the elderly. The end goal of autonomous cars can help to reduce accidents over time. The elderly population growth is flat, but the number of accidents for those over 60 is growing. The intent of this latest project is to provide the driving capabilities of an expert driver to the elderly population.
The autonomous car is one of the fundamental technologies for the expert driver project. One key technology is data acquisition. An expert driver uses a 10 second horizon, with a window at 5 seconds to predict hidden risks. Emergency avoidance occurs in the last second. The tasks for the expert driver are to plan a path, determine risk potential, and avoidance control. One challenge is to create a user interface that allows for ease of use and adoption while controlling automatic braking and steering.
A scenario database of likely situations can assist the risk potential control with various risk profiles. This expert database has to integrate defensive driving behaviors while helping to reduce human error. The user interface has to consider three control paths; sensor data, data fusion, and big data environmental sensing. The data have to lead to sense and recognition functions with high reliability through high performance compute and integrated databases for full car control.
Due to the constantly changing conditions, the intelligent car needs to access both local and global driving intelligence. Integrated communications to all other smart cars augments the static database and the other sensor data. An example of a partial solution is the adaptive cruise control that changes speed to adjust to traffic and terrain conditions. Adding automatic merge capabilities and supplementing the whole area with signage will help to include the uninstrumented cars.
Vehicle to vehicle communications operating at 750 MHz and mm wave radar for vehicle spacing will allow platooning and is a fairly small addition to the lane following technology. One issue is how to alert the occupants of a needed driver action without startling the driver and causing panic and inaction. The dynamic nature of the traffic environment is a high hurdle for the autonomous car.
The basic technologies for the smart car will also require other infrastructure enhancements. The cars will need some standard communications systems and protocols for data sharing, and the extended global databases will have to have very low latency to be of value to the driving public.
