자율주행 대중교통 차량의 공사 영역 침범 방지를 위한 인지 기술

초록

The problem of work zone intrusion, in which a vehicle invades the interior of a road construction site, has been raised before the introduction of autonomous driving technology. Existing research was a method of installing sensors around the work site and detecting and notifying vehicles infiltrating the site. Although it is best when the vehicle cannot be controlled, the development of a cognitive system for autonomous vehicles can fundamentally prevent intrusion of the work area. Accidents involving intrusion into the work area can be prevented through the technology that the recognition system of the autonomous vehicle recognizes as an area that includes all of the individual objects to be recognized. In this study, we propose a cognitive technology that finds the outermost boundary of the work area that includes all obj ects detected at the road construction site, and considers it as one big obstacle to prevent the intrusion of the work area. Based on the Occupancy Grid Map, we formulate the problem of finding an area containing all objects and displaying objects. This is a problem of finding a point constituting a convex hull when there are multiple points, and there is a problem in that the amount of calculation increases as the size of the grid decreases. In this study, the amount of computation is reduced by selecting a representative grid for objects represented by multiple grids to find a convex hull. The operation of the proposed algorithm is verified through field experiments, and the calculation amount is reduced based on field data.

키워드