Roadworks can be hazardous for both road workers and road users. Even with state-of-the-art safety measures in place, serious accidents do happen there, particularly when drivers do not heed roadwork signs and speed limits. Crashes at roadworks that involve killed or seriously injured (KSI) casualties account for about 2% of all KSI crashes in developed countries, even though the roadworks are normally well-signaled and are also marked in quick-reaction road/traffic maps. These media provide several means for the drivers ‒ and for the on-board advanced driving assistance systems (ADAS) helping them ‒ to duly detect roadworks. In the paper, an approach based on statistical inference is presented for detecting roadwork zones. The approach takes into account the engineering regulations and practice concerning setting up temporary road configurations near and along roadworks. Such configurations often involve narrower traffic lanes and traffic signs installed closer to traffic. The approach detects change in ‒ among other type of collected data ‒ the lateral positions of the traffic signs measured relative to the ego-car along the road. In a practical implementation, the traffic sign detection and recognition, the lateral distance measurement and the data recording are carried out by some traffic sign recognition (TSR) system. The traffic sign data is seen as a realization of a marked Poisson process and the minimum description length (MDL) principle – set to work in the form of Page-Hinkley change detectors – is applied for detecting roadworks.
Publication Date: 2017
Full Text URL: Link to URL
Publication Types: Books, Reports, Papers, and Research Articles
Topics: Crash Prevention; Data Collection; Detectors; Driver Support Systems; Signs; Work Zones