Development of a tool for rider posture tracking on a motorcycle

Safety has become one of the most important fields to develop for the vehicle industry. Both in the automotive industry and others, new features for traditional vehicles are being developed to increase the safety and comfort of passengers, such as autonomous driving. In the case of the motorcycle industry, autonomous driving is further away from implementation, so this sector is developing other ways to increase the safety of its vehicles. At the Technische Universität Darmstadt, a test motorcycle is setup to better understand how motorcycle riders behave in different situations. The system is based on different sensors and cameras to capture the status of both the motorcycle and the rider in each of these situations. One of the objectives of this project is to develop a tool that fits the TU Darmstadt project to estimate the position of the rider through fisheye cameras located on the back and front of the motorcycle. For this purpose, different methods of object detection through image processing have been evaluated, taking into account the potential of each and the final objective of this tool. The results of these different tests have resulted in a system composed of small markers that, incorporated on the jacket of the rider, allow to extract the position of each one of the locations. For this, a tool based on three stages has been developed. The first is to extract the parameters of the camera to be able to analyse the images afterwards. The second consists of compensating the distortion caused by the fisheye cameras, using the parameters extracted in the first stage. In the last stage, the tool allows to analyse the position and orientation of the markers in the image and converting it into the position it occupies in the motorcycle coordinates. This tool has been tested with different tests to analyse its accuracy. Analysing the results, it is determined that this tool is a valid approach to satisfy the purposes of the project of the Technische Univertität Darmstadt. The precision, although variant in each of the tests, is sufficiently high to satisfy its objective in the ranges in which the tool has been configured. Finally the tool is tested under the conditions for which it was designed, and the possible aspects to be improved are detailed for its subsequent implementation.Outgoin