|M.Sc Student||Bolshinsky Ella|
|Subject||Towards an Autonomous Sailing Trainer|
|Department||Department of Computer Science||Supervisor||PROF. Roy Friedman|
|Full Thesis text|
This work reports on our experience in designing and implementing a software/hardware bundle that implements a technical sailing trainer for the sailors of a Laser dinghy sailboat. The goal of such a trainer is to enable sailors, especially at the youth level, to practice in a genuine environment even when a human trainer is not available. Trying to keep the trainer simple and affordable, the bundle is composed of a smartphone and a small set of Wi-Fi enabled cameras placed in designated places on the boat. Computationally expensive tasks may be processed on a remote server. The bundle analyzes the condition of the sailboat by processing the pictures taken by the cameras and the smartphone sensors' reading and provides instructions to the sailor based on this analysis. The recommendations are generated from a combination of heuristics and machine learning mechanisms. Some of the main challenges of the system include extreme changes in lighting and weather conditions, sea spray, slight shifts in cameras locations due to strong winds and the waves, tracking multiple moving objects, and terminating all the processing in real-time before the next set of images and sensor readings arrives. A prototype of the bundle was created and implemented. The performance and limitations of all components were studied, although due to technical barriers, we fell short of a complete live deployment. In the process, we have accumulated some interesting insights about building such low cost real-life autonomous systems.