T-Timespan Dynamic Coverage Of Mobile Camera Networks

The coverage problem based on camera networks is a popular topic in acdemic research.Visual inspection is one of the applications of coverage problem,playing an improtant role in industry and daily life.From the perspective of whether camera networks move during the task,the coverage problem can be classified into static coverage and dynamic coverage.The goal of static coverage is to improve the coverage performance by optimizing the fixed positions and orientations of camera networks.However,visual inspection task requires continuous cameras’ movement which static coverage cannot meet.Therefore,this paper focuses on the dynamic coverage problem of camera networks.Due to the anisotropy of cameras in practical applications,it has a significant impact on the coverage effect.Therefore,the coverage model constructed in this paper is the projection of a three-dimensional frustum onto a two-dimensional plane,which preserves this characteristic.In order to evaluate the coverage quality of the information captured by cameras,this paper introduces the concept of homomorphic transformation and defines the coverage strength by calculating the homomorphic distance.In addition,this paper considers the sustainability of coverage and proposes the concept of T-timespan.Specifically,when a task point is covered at time t,the covered state will last for T seconds,and then becomes invalid after time t+T waiting for the next coverage by the camera.The value of T can be determined according to users’ requirements or task demands.This paper proposes the T-timespan dynamic coverage problem of mobile camera networks and the main content of this paper is as follows:1.As camera is an important tools for visual inspection,the traditional two-dimensional model lacks the realism of three-dimensional object.This paper projects a three-dimensional field of view model onto a two-dimensional plane to construct a coverage model that can be directly applied to camera coverage tasks in 3D.The coverage model is further subjected to homomorphic transformation to reduce the computational complexity caused by anisotropy,and the homomorphic distance is calculated based on the Euclidean distance to design the coverage intensity as an evaluation criterion.2.This paper designs a general dynamic coverage cost function based on the T-timespan,which can solve different coverage problems by changing the value of T,and is verified through simulations.3.Considering the hardware conditions and structural limitations of intelligent vehicles,this paper introduces a kinematic model with nonholonomic constraints,and considers this model when using the gradient ascent algorithm to optimize the cost function.The effectiveness of the control law is verified through physical experiments.