Robo-Yacht: a human behaviour-based tool to predict the performances of yacht-crew systems

Many numerical tools are available in the yacht design domain to predict yacht performances. These are based on experimental data and well-established numerical techniques. However, being competitive sailing an uncertainty-rich discipline, there is also a need to assess the performances of a yacht’s crew. This is indeed the main area for performance gains. The present study aims at predicting the performances of yacht-crew systems, by including numerical models for human behaviour within those referred to the yacht dynamics.In particular, the problem of decision-making under weather uncertainty is formulated in terms of a game of chance having nature as a second player and involving risk. Within this context, it is shown that decisionmaking models often used in management sciences can be advantageously used. This approach has led to the development of a sailing simulator referred to as ‘Robo-Yacht’, based on the International America’s Cup Class. A case study is investigated that involves three strategical alternatives and four possible weather scenarios: gains and losses are assessed through the simulator and a formula to express expected payoffs is derived. When different attitudes towards risk (neutral, risk-averse and risk-taking) are expressed as different utility functions,it is shown that sailors’ choices can be conveniently modeled based on a maximization of expected utility. The ‘automatic crew’ can actually make decisions that appear to be consistent with widely accepted principles of race strategy. A risk function is also derived in order to take into account opponents’ choices and, where necessary, modify the strategic plan accordingly.