Automatic Detection of Stains on Lidar Glass Houses and Notice for Cleaning
Loading...
Authors
Toshniwal, Shradha
Issue Date
2021
Type
Thesis
Language
Keywords
Lidar , Offset , Sensor , Stains , Threshold
Alternative Title
Abstract
For achieving a sustainable and smarter transportation system, sensor technology needs to be combined with transportation infrastructure. Traffic sensors are significant in today’s world since the conventional visual inspection is inadequate for steering quality control and traffic safety, efficiency being of utmost importance, high-speed and accuracy automated inspection becomes crucial. No system is perfect, and Lidar is no different.Although Lidar, sensor has gained its popularity with its 360-degree monitoring and visualization, being a relatively new technology, it has its frail spots too. Mainly, on roadside, factors like surface obstacles or environmental condition, influence its performance with uncertainty of cloud point movement from its true value. Therefore, through this study, a standard method based on the difference of offset is proposed to check the quality of data for real road deployment and answer a very foundation question from traffic engineering, about the obstacles recognition on lidar glass houses, and how often to clean the sensor increasing confidence on such systems . In this study, multiple experiments, comparing different conditions of sensor surface was conducted where real time frame was compared to standard frame and the frame offset was used to define a threshold value and over threshold offset time. The experiment was conducted with varied beams, scenarios and further the method was validated with real time traffic data. Sensor itself may have many cavities; we only needed to know the limiting range to accommodate real traffic. This study therefore contributed by developing method to find out influence of stain on sensor automatically and notify related agency, the time to clean the sensor without validation from engineers at intersections at every small interval.