IPOD Abstract for presentation (Poster or Podium) with a Paper in the Conference Proceedings
Rail Transport
Pan Lu, PhD
Associate Professor
North Dakota State University
Fargo, ND, United States
Pan Lu, PhD
Associate Professor
North Dakota State University
Fargo, ND, United States
Yihao Ren, PhD
Assistant Professor
University of North Carolina at Pembroke
Fargo, ND, United States
Yihao Ren, PhD
Assistant Professor
University of North Carolina at Pembroke
Fargo, ND, United States
Chengbo Ai, PhD
Assistant Professor
University of Massachusetts Amherst
Amherst, MA, United States
Chengbo Ai, PhD
Assistant Professor
University of Massachusetts Amherst
Amherst, MA, United States
Xinyi Yang, MS
Research Assistant
North Dakota State University
FARGO, ND, United States
Xinyi Yang, MS
Research Assistant
North Dakota State University
FARGO, ND, United States
Denver Tolliver
United States
Denver Tolliver
United States
Xianfeng Yang, n/a
assistant professor
University of Maryland
College Park, Maryland, United States
Xianfeng Yang, n/a
assistant professor
University of Maryland
College Park, Maryland, United States
Pan Lu, PhD
Associate Professor
North Dakota State University
Fargo, ND, United States
Shortlines are smaller railroads that operate over shorter and light-density rail lines. The track conditions of them are often marginal, their revenue is insufficient, and their number of employees is declining. Shortlines depend on labor-intensive yet timely inspections to ensure their infrastructures are safe and effective to operate. To increase track geometry inspection efficiency, research efforts have heavily focused on developing automatic and effective rail extraction methods using LiDAR and artificial intelligence (AI) technologies, so that track geometries can be calculated based on the extracted 3-D rail model. However, many of these methods were developed with specific requirements by Class I railroads and have poor adaptability when deployed for shortlines. This research will collect additional shortline track data using LiDAR sensor and such data will be used to validate and assess the performance of an AI-assisted rail extraction method for LiDAR point cloud data. The original method was designed using a set of publicly available data published by the Federal Railroad Administration and wasn’t evaluated with external data set especially shoreline data.