Inland waterway transportation is one of the most important means to transport cargo in rivers and canals. To facilitate autonomous navigation for ships in inland waterways, this paper proposes a data-driven approach for predictions and control of underactuated ships with unknown dynamics, which integrates model predictive control (MPC) with an iterative learning control (ILC) scheme. In each iteration, kernel-based linear regressors are used to identify the relations between the evolution of ship states and control inputs based on the stored data from previous iterations and the collected data during operation, so as to build the system prediction model. The data are dynamically used to fix the prediction model over iterations, as well as to improve the controller performance until it converges. The proposed approach does not require prior knowledge regarding the hydrodynamic coefficients and ship parameters, but learns from the data instead. In addition, it exploits the advantages of MPC in handling constraints with minimised overall cost. Simulation results show that the controller could start from a nominal, linear data-driven ship model and then learn to reduce the path-following errors based on the data obtained over iterations.

In recent years, several frameworks and systems have been proposed that extend Inductive Logic Programming (ILP) to the Answer Set Programming (ASP) paradigm. In ILP, examples must all be explained by a hypothesis together with a given background knowledge. In existing systems, the background knowledge is the same for all examples; however, examples may be context-dependent. This means that some examples should be explained in the context of some information, whereas others should be explained in different contexts. In this paper, we capture this notion and present a context-dependent extension of the Learning from Ordered Answer Sets framework. In this extension, contexts can be used to further structure the background knowledge. We then propose a new iterative algorithm, ILASP2i, which exploits this feature to scale up the existing ILASP2 system to learning tasks with large numbers of examples. We demonstrate the gain in scalability by applying both algorithms to various learning tasks. Our results show that, compared to ILASP2, the newly proposed ILASP2i system can be two orders of magnitude faster and use two orders of magnitude less memory, whilst preserving the same average accuracy.