Streamlined A* for Faster Robotic Inspections in Ports

WARDANA, Hartanto and RUMAKASARI, Atyanta Nika and PICANUSSA, Prischa Wilhelmina and SOOAI, Adri (2024) Streamlined A* for Faster Robotic Inspections in Ports. Jurnal Sistem Cerdas, 7 (2). pp. 175-188. ISSN 2622-8254

Text Streamlined A_ for Faster Robotic Inspections in Ports.pdf Restricted to Repository staff only Download (1MB)

Abstract

Research on automatic port inspections using robots has been carried out in the state-ownedcompany Indonesia Port Corporation, Semarang Indonesia. However, increasing the efficiency of robotic inspections is critical because robots need to perform these tasks with much higher speeds than humans, while maintaining a high level of accuracy. The robot is equipped with sensors and computer vision technology to detect defects or problems that the human might miss. This aim is to increase overall inspection accuracy at a lower cost. In this research, introducing an optimized A* path planning algorithm that incorporating with the flood algorithm, node reductions process, and linear path planning optimization for an autonomous navigated port inspection robot. Our primary objective is to significantly increase the efficiency of the conventional A* algorithm in guiding robotic systems through complex paths. The proposed algorithm demonstrates exceptional efficiency in generating feasible paths, with success attributed to optimization steps that specifically target reducing node processing and enhancing route finding. The experimentation phase involves a comprehensive assessment of the algorithm using six key parameters: running time, number of nodes, number of turns, maximum turning angle, expansion nodes, and the total distances output. Through rigorous testing, the algorithm's performance is evaluated and compared against seven other current algorithms, namely A*, BestFirst, Dijkstra, BFS, DFS, Bidirectional A*, and Geometric A*. Results from the experiments reveal the algorithm's outstanding running time efficiency, surpassing all other algorithms tested. Notably, it exhibits a remarkable 6.5% improvement over the widely recognized Geometric A* algorithm.

Actions (login required)

View Item