Expert Systems with Applications
The operation of a ship in the ocean depends crucially on the quality of routine offshore dry dock maintenance. Automation by robotics is an efficient solution to address the issues of saving water, energy, time, and easing the labour workload when conducting hydro-blasting hulls in the dry dock ship maintenance industry. In this paper, the automated hydro-blasting in corroded ship hull cleaning by a novel robot platform with reconfigurable manipulators named Hornbill is proposed. The robot is able to maneuver smoothly on a vertical surface by permanent magnetic force, to carry the heavy load, to clean the corroded ship hull by hydro-blasting, and to self-evaluate hydro-blasting task by leveraging the Deep Convolutional Neural Network (DCNN) to synthesis the corrosion level map of the blasted workspace. We also propose an optimal complete waypoint path planning (CWPP) framework to help the robot re-blast the benchmarked workspace. The optimal CWPP problem, including objective functions of the shortest travel distance, the least upward moving direction to reduce water, energy spent while ensuring the visiting of the robot to all uncleaned waypoints defined by benchmarking output, is modeled as the classic Travel Salesman Problem (TSP). The evolutionary-based optimization techniques, including Genetic Algorithm (GA) and Ant Colony Optimization (ACO), are explored to derive the Pareto-optima solution for given TSP. The experimental results show that the magnetic force and motors torque are synchronized to enable the proposed system to navigate smoothly on the vertical surfaces tested with different corrosion levels. The proposed corrosion level benchmarking achieves a mean accuracy of 0.956 with an execution time of 30 fps. Besides, the proposed CWPP enables the proposed robot to yield about 15%, 26%, and 5% the energy, water, and time, respectively, less than the conventional methods when the experiments are conducted in various workspaces on the real ship hull.