Third Rail Wear Inspection Based on Point Cloud Registration

摘要

Third rail inspection is vital to ensure that the train daily operations are able to run smoothly. With the limited maintenance downtime during the wee hours for inspection and maintenance, it is imperative to improve the time required while ensuring the maintenance crew safety. In this paper, singular value decomposition - Iterative Closest Point is proposed for the inspection of surface wear on the third rail. Firstly, the profile of the third rail data is acquired by a line laser. After the data had been acquired, unwanted data is removed and the measured model is split into two parts for registration. The measured model is then registered to the reference model. Finally, the wear amount is analyzed using the Hausdorff distance.