Research on a Train Safety Driving Method Based on Fusion of an Incremental Clustering Algorithm and Lightweight...

Area of Science:

• Computer Science
• Robotics
• Transportation Engineering

Background:

• Railway transportation systems face significant safety risks from unknown obstacles on tracks.
• Current obstacle detection methods often struggle with accuracy and real-time processing for unforeseen objects.

Purpose of the Study:

• To develop an innovative and reliable obstacle detection strategy for railway tracks.
• To improve the accuracy and real-time performance of identifying unknown obstacles.

Main Methods:

• An incremental clustering algorithm is employed to merge point cloud clusters and eliminate noise.
• Dilation and erosion theories are used to refine cluster boundaries.
• Lightweight shared convolutional semantic segmentation algorithms are integrated for precise obstacle localization.

Main Results:

• The proposed method achieves an average obstacle detection recall rate of 90.3% on public datasets.
• The strategy effectively identifies and connects spatially adjacent point cloud clusters.
• Noise reduction in target object point clouds enhances recognition precision.

Conclusions:

• The integrated detection strategy significantly improves the accuracy and real-time capabilities of railway obstacle recognition.
• The method demonstrates high reliability and practical application value for ensuring railway safety.
• This approach offers a robust solution for detecting unforeseen obstacles in railway transportation.