Mounted Machine Learning Camera System for Object Detection and Track Tracing During Hyperloop Operations

ABSTRACT

This research focuses on addressing safety concerns in designing a cost-effective and space-efficient Hyperloop system. The main safety factors include managing debris, cracks, and defects in the tunnel. To achieve this, the study explores the integration of sensors and equipment while avoiding excessive costs and space consumption.

The motivation behind this research lies in two essential factors. Firstly, to develop an advanced safety system for the Hyperloop pod that ensures secure transit by detecting faults, obstacles, and faulty sensors. These issues pose significant risks and potential destruction to the pod during transit if not effectively addressed. Secondly, the aim is to enhance quality assurance throughout the system.

The research culminates in the creation of a camera system capable of identifying objects and tracking Hyperloop tracks. This system incorporates mechanical and hardware designs suitable for mounting on experimental Hyperloop pods. Moreover, the software developed for this purpose efficiently works with the physical components to meet the Minimum Viable Product (MVP) requirements. The camera system utilizes the YOLO algorithm and Python libraries, such as TensorFlow, to achieve the desired functionalities outlined in the research motivation. Multiple datasets were generated and used to train the machine-learning model, enabling the system to identify various common items as a proof-of-concept.

Notably, the camera system has been designed with 3D printing in mind to streamline the manufacturing process and relies predominantly on off-the-shelf hardware components. The software side employs Python programming to develop algorithms necessary for training and utilizing machine-learning models. Overall, this research successfully presents a viable and efficient solution to address safety concerns in the Hyperloop system while maintaining cost-effectiveness and minimal space consumption.