面向协同感知的任务分配方法

Collaborative Awareness Oriented Task Allocation Method

为维护社会公共安全,当出现违法案件时,公安部门需尽快获取案件发展的实时信息。目前实时获取信息的渠道主要有两种:一是利用部署在城市热点地区大量监控摄像头进行追踪,二是利用普通民众携带的各类智能设备获取信息。因此提出一个面向协同感知的任务分配方法,利用监控摄像头与行人通过协同感知覆盖特定区域道路网络。首先根据实验区域内道路网络拓扑结构研究任务重要程度,提出任务优先级计算方法,用以区分不同任务优先级;其次提出参与者可信度计算方法,该方法以物理距离为基准,通过引入空间距离衰减指数函数计算行人执行某项待选任务的意愿值,再与任务优先级结合,计算行人可信度;最后提出双向选择多轮任务分配(McTA)算法并利用真实数据集进行实验测试。结果表明,该方法在数据集中完成率达95%以上。

In order to maintain the public security,when there are illegal cases,the public security department needs to obtain the real-time information of the case development as soon as possible.For this purpose,two major approaches are usually adopted.One is to use a surveillance network consisting of a large number of surveillance cameras deployed in urban hotspots;the other is to use various smart devices carried by people to obtain information.Here we propose a task allocation scheme for collaborative sensing.The method aims to make the cameras and pedestrian to jointly cover the road network of the target area.We first calculate the importance value of each task,which can be used for prioritizing a bunch of tasks.Then we propose a method to evaluate the credibility of participants.This method is based on Euclidean distance while involving a spatial distance attenuation function to compute the pedestrians’willingness and combining with the task priority.Lastly,we propose mutual-choice task allocation(McTA)algorithm.The extensive evaluation on real dataset demonstrates that the proposed scheme can achieve over 95%task allocation rate.