Autonomous Electric Boat Collision Avoidance Based on Artificial Potential Field

This thesis presents the development and evaluation of an innovative collision avoidance algorithm for an Unmanned Surface Vehicle (USV) designed for cruising the canals of Venice. The study focuses on the development of a motion control based on Artificial Potential Field (APF) theory with the use of local attractor able to guide the USV detouring through preferred regions. Further, it is compared to a Nonlinear Model Predictive Control (NMPC) algorithm integrated with Dubins Curves based Rapidly-exploring Randoms Trees* (RTT*) for path planning. Both aim to ensure safe and efficient navigation in changing marine environment adjusting USV trajectory on the basis of static and dynamic obstacles perceiving. On the other hand, a further goal of this study is evaluating an autonomous navigation method able to guarantee ColReg rules compliance. Through extensive simulation conducted on Matlab software the APF-based method fulfils this task for both static and dynamic obstacle representing a valid alternative to other techniques. In the conclusion it is presented a comparation of the results arisen from these two approaches.