Sensor fusion is the basis for future development of automated driving. Especially the fusion of millimeter wave radar and camera has not only enhanced the precision and accuracy of the target detection, but also reduced the false warning rate. In this way, it is able to greatly enhance the practicability of the FCW (forward collusion warning), AEB (automatic emergency braking) and ACC (adaptive cruise control) as well as other ADAS system functions.
In the evolution from L1 to higher level of automated driving, the coordination and fusion among several sensors contributes to guarantee the safety of system performance and functions. Different sensors have different advantages and disadvantages. Currently, not yet a kind of sensor could apply to all kinds of operating environment. So only through the fusion of sensors, can the “robustness” or diversification of functions be more appropriately realized.
Sensor fusion is mainly divided into the following levels according to different strategies: image data level (filter of the millimeter wave radar with the camera as the main sensor), target feature level (fusion of the target output of camera and radar with the millimeter wave radar as the main sensor), and decision level (use of VS algorithm comparison to make decisions).