기술의 발전으로 스마트 선박과 관련된 다양한 연구가 진행되고 있으며, 기관실을 무인으로 순찰할 수 있는 기관실 순찰 로봇 도 이러한 연구 중의 하나이다. 순찰로봇은 인공지능을 통해 학습된 정보를 기반으로 기관실을 이동하며 기기 정상 유무 및 누수, 누유, 화재 등의 이상 유무를 파악한다. 기관실 순찰로봇에 관한 연구는 인공지능을 이용한 객체 검출에 관한 연구가 주로 진행되고 있으나, 순 찰로봇의 이동 및 제어에 관한 연구는 부족한 상황이다. 이는 순찰로봇이 객체를 검출하더라도 검출한 객체까지 이동할 방법이 없다는 문제를 야기한다. 이에 본 논문에서는 기관실 이상상황 발생 시 빠르게 이상 유무를 파악할 수 있는 기동성을 확보하기 위해, A* 알고리 즘을 적용하여 순찰로봇이 최단경로를 탐색할 수 있는지를 확인하였다. 라이다를 장착한 소형차를 이용하여 선박 기관실을 주행하며 데 이터를 얻어, SLAM으로 매핑하여 지도를 만들었다. 매핑한 지도에서 순찰로봇의 출발 지점과 목표 지점을 설정하고, A* 알고리즘을 적용 하여 출발 지점부터 목표 지점까지 최단 경로를 탐색하는지를 확인하였다. 시뮬레이션 결과 매핑된 지도에서 출발 지점부터 목표 지점까 지의 장애물을 회피하며 최단 경로를 잘 탐색함을 확인 할 수 있었으며, 기관실 순찰로봇에 적용하면 선박안전에 도움이 될 것으로 사료 된다.

The shortest path problem is one of network optimization problems. This paper considers a shortest path problem under the situation where lengths of arcs in a network include both uncertainty and randomness, and focuses on the case that the lengths of arcs are expressed by uncertain random variables. This paper presents a new type of model: relative entropy model of shortest path. By the definition of relative entropy of the uncertain random variables, relative entropy model of shortest path problem is proposed to find the shortest path which fully reflects uncertain and random information. This model is formulated to find a shortest path whose chance distribution minimizes the difference from the ideal one. A numerical example is given to illustrate the model’s effectiveness.

We propose a new algorithm for a classical problem in the planer computational geometry: computing a shortest path between two points in the presence of circular obstacles. Proposed algorithm actually computes a path tree that encodes a shortest path between given two points. Types of path are defined as a tangent line segment between circles, between point and circle, or as an arc. Using circle set voronoi diagram, geometric information that is very useful to search circles is obtained. The key of proposed algorithm is the reduction of the number of circles to need for constructing path tree. The main advantages of our algorithm are its robustness, speed, and the simplicity in implementation.

We give efficient algorithms for finding shortest distances in special types of sparse network where the networks are composed of series of subnetworks which form circles. The algorithms presented in this paper are so called extensions of the non-circular sparse network algorithms The suggested algorithm is composed of two approach parts for only donut shaped sparse network, the one is the node elimination approach and the other the arc elimination approach, Our algorithms for finding all shortest distances can be extended in general circular sparse networks.

A* algorithm is a global path generation algorithm, and typically create a path using only the distance information. Therefore along the path, a moving vehicle is usually not be considered by driving characteristics. Deceleration at the corner is one of the driving characteristics of the vehicle. In this paper, considering this characteristic, a new evaluation function based path algorithm is proposed to decrease the number of driving path corner, in order to reduce the driving cost, such as driving time, fuel consumption and so on. Also the potential field method is applied for driving of UGV, which is robust against static and dynamic obstacle environment during following the generated path of the mobile robot under. The driving time and path following test was occurred by experiments based on a pseudo UGV, mobile robot in downscaled UGV’s maximum and driving speed in corner. The experiment results were confirmed that the driving time by the proposed algorithm was decreased comparing with the results from A* algorithm.