Testing of SMRT train door system (object detection and recognition using OpenCV on Raspberry Pi)

Abstract

Unnecessary malfunction and breakdown in trains’ causes inconvenience to passengers and affect service intervals. One commonly found fault in train systems are the door faults. In the interest of safety, passengers are detrained whenever a door fault occurs or doors that cannot be closed. In events that these happens, the train would be withdrawn from service and send back to train depot for servicing. One of the challenges faced by train captains and maintenance crews are that not all door faults are technical in nature, as it could be cause by non-technical events such as passenger behavior or by foreign objects trapped at doorway, preventing door from closing. These faults are difficult to detect and tedious to discover. It is important that a real-time condition monitoring system can be used to alert train captains or maintenance crew and assist them in identifying the fault so as to facilitate quicker troubleshooting procedures. A condition monitoring system using raspberry pi would be recommended due to its low power consumption and relatively small size. Furthermore, raspberry pi is able to run python programming platform remotely which would be ideal because it will be installed on all train doors.