面向自动协同驾驶的多车编队任务分配策略

自动驾驶的实现需要大量车载传感器的支持，然而，在有限车载计算资源条件下，由传感器所产生的庞大数据量使得自动驾驶任务的实时性难以满足，成为阻碍自动驾驶技术进一步发展的重要阻力。通过将驾驶任务进行协作处理，因而充分利用多个协作车辆的计算资源，自动协同驾驶成为解决该问题的新途径。而如何形成多车编队并实现编队中驾驶任务分配则是实现自动协同驾驶的关键。该文首先采用排队理论G/G/1模型建立一种普适性车辆编队网络拓扑分析模型，充分考虑编队内车辆间的任务协作能力和单个车辆的任务负荷，得出任务的处理时延和车辆系统中的平均任务数；其次，采用支持向量机(SVM)方法，基于车辆的负荷程度及处理能力将车辆的“空闲”、“繁忙”两状态进行分类，进而建立针对车辆协作任务分配的候选车辆集。最后，基于上述分析，该文提出面向多车编队协同驾驶的任务均衡策略——基于分类的贪婪均衡策略(C-GBS)，以充分平衡编队内所有车辆的任务负荷并利用不同车辆的任务处理能力。仿真结果表明，该策略能够减小重负荷网络中的任务处理时延，有效提升自动驾驶车辆的任务处理效率。

Task Assignment Strategy for Platoons in Cooperative Driving

Autonomous vehicles are equipped with multiple on-board sensors to achieve self-driving functions. However, a tremendous amount of data is generated by autonomous vehicles, which significantly challenges the real-time task processing. Through multiple-vehicle cooperation, which makes the best of vehicle onboard computing resources, autonomous and cooperative driving becomes a promising candidate to solve the aforementioned problem. In this case, it is vital for autonomous and cooperative driving to form a driving platoon and allocate driving tasks efficiently. In this paper, a more general analytical model is developed based on G/G/1 queueing theory to model the topology of platoons. Next, Support Vector Machine (SVM) method is adopted to classify the “idle” and “busy” categories of the vehicles in the platoon based on their computing load and task processing capacity. Finally, based on the analysis above, an efficient task balancing strategy of platoons in autonomous and cooperative driving called Classification based Greed Balancing Strategy (C-GBS) is proposed, in order to balance the task burden among vehicles and cooperate more efficiently. Extensive simulations demonstrate that the proposed technique can reduce the processing delay of driving tasks in platoons with high computing load, which will improve the processing efficiency in autonomous vehicles.