Safety Awareness Online Detection System of Driving Behavior Based on Software and Hardware Co-design

The safety awareness online detection of driving behavior means that a kind of detection of the ongoing unsafe driving behavior as it starts but before it ends. This type of detection is significant important for achieving safe driving assistant because it is able to recognize the unsafe or aggressive driving behavior in advance and then give a buffer time to issue the vital alert. In order to achieve online detection, it is critical to reduce the processing time spent on the whole behavior recognition procedure, which usually is complex and involves a lot of computation loads. This paper proposes an approach to address the challenge by adopting a methodology of software and hardware co-design. We implement the most computation intensive stages of preprocessing and feature extraction as the hardware cores. A prototype safety awareness online behavior detection system is implemented on the programmable chip (SoPC) by using software and hardware co-design, and the experiment shows its effectiveness.