Research And Implementation Of Platoon Cooperative Context Sensing System

With the development of artificial intelligence in recent years,the automobile industry is developing towards connection and intelligence.It is very important to improve the ability of traffic context sensing for driving safety.Nowadays,the assisted-driving devices in the market almost need special hardware,and their prices are too expensive to afford for most people.On the other hand,the android phones are becoming more and more functional powerful,so that they can handle complex computing tasks.What's more,with the portability and the low prices,the android phones will be widely applied in intelligent connected vehicle.In this paper,a platoon cooperative context sensing system is researched and implemented.In the system,image acquisition,image transmission,image stitching and object detection are realized by android phones and OBUs,which expands the drivers',perception of the surrounding context.While improving the driving safety,the system also contributes to save energy and reduce exhaust emissions.Firstly,this paper proposes an image acquisition method based on context sensing.The appropriate image resolution is selected according to the current context.Although the accuracy is slightly reduced,there is a significant decrease for system delay and resource consumption,achieving a trade-off between accuracy and real time.Secondly,an image transmission platform based on vehicle-to-vehicle communication is set up.The platoon is constructed with android phones and OBUs to realize the image transmission among android phones.Thus,the drivers'vision is enlarged,which improves the ability of context sensing greatly.Thirdly,the distributed computing method is applied for image stitching and object detection in the platoon.The corresponding task segmentation rules and collaborative computing rules are designed respectively.Thus,the whole computing task is divided for every device in the platoon to reduce the system delay and resource consumption.There are still wide areas to study for the proposed system in this paper.For example,the packet loss should be reduced further to improve the accuracy in the image transmission.In addition,for the collaborative distributed computing,a more appropriate image segmentation rule that considers the processing capabilities of all nodes should be adopted to improve the system performance.