F2008-01-011
A Situational Awareness Architecture for Vehicle-to-Vehicle Communication
Vehicle-to-Vehicle communication (V2V) introduces a new mechanism for vehicles to share information with each other. Given the relatively low-latency communication compared to other devices such as cell phones, the information shared via V2V can be used to support safety and safety enhancing use cases. This essentially introduces a new "sensor" into the vehicle that allows one vehicle to "detect" other equipped vehicles within some distance. However, once vehicles can communicate with each other, what information should they share and how should they process all of it? Past work in various public projects has identified what information should be shared and resulted in awareness messages with elements such as speed, acceleration, yaw-rate, and location. This paper presents a new method for processing the information received and feeding intermediate results to algorithms supporting various use cases.
First, the paper presents various requirements for different use cases and proposes a Situational Awareness functional block to pre-process information. The use case requirements reveal a common need for classification of neighboring vehicles relative to the host vehicle. To minimize redundant processing between multiple use case algorithms, we propose a functional block named Situational Awareness to classify the incoming data before providing it to any use case algorithm. The Situational Awareness processing starts upon the reception of any new wireless situational awareness message and classifies the location of the sending vehicle. The use cases relevant to that classification then process the data with the use case-specific algorithms.
Next, the paper proposes a unique architecture for the newly defined Situational Awareness functional block. At a basic level, Situational Awareness uses the past driving "trail" of the host vehicle as well as other vehicles to create a base classification for another vehicle as either Oncoming, Intersecting, Same Road Ahead, or Same Road Behind. When highly accurate positional data is available and other vehicles are in a "Same Road" classification, Situational Awareness classifies the relative lanes of the other vehicles compared to the host vehicle. After the classification, the use case algorithms process vehicle data for vehicles with an appropriate classification. For instance, a Forward Collision Warning algorithm would process only information from vehicles classified as Same Lane Ahead. In addition to the classification, the architecture provides an innovative mechanism to prioritize processing of vehicles that are of most importance to the use cases. The prioritization is most important in situations where hundreds or even thousands of vehicles are within communication range and the processing power is limited.
Finally, the paper presents real-world performance data for road and lane classification relative to operation on highways, in rural areas, and in city environments. The data shows that in many environments the classification algorithm is extremely reliable. However, there are limitations that correspond to the accuracy and availability of GPS in environments with large buildings and trees. In future deployments, the limitations must be identified in real-time by both the Situational Awareness algorithm and by the use case algorithms.
Poster presentation: Mobility concepts