Mining truck activity recognition

Activity recognition of mining trucks is a challenging problem for human designed algorithms due to the large variety in vehicle movements that could represent a given mining action.

Human designed algorithms don’t generalise easily, and always have edge cases. Deep neural networks have been shown to provide substantial improvements over shallow classifiers for activity classification in other research fields. 

This project aims to apply deep learning methods to the problem of activity recognition of heavy mining vehicles, with the ultimate aim of exploring their real world effectiveness for automated feature extraction and activity classification on mining trucks. The model receives the truck suspension data, gear position and speed output to predict whether the truck is the process of loading or dumping ore.

Theme
Transforming technologies

Booth
TT55

School
Computer Science

Exhibitor
Adrian Orenstein

vote for this project: TT55

Back to project list