Mobile sensors' patrol path planning for a border surveillance application

Name: Wichai Pawgasame

keyword: Path planning

; Automated planning

; Mobile sensor network

; Planning in unobservable environment

; Planning in partially observable environment

; Border surveillance

Abstract: The border control becomes very challenging, when a limited number of border patrols is deployed in a large border region. To compensate for the limited number of border patrols, the border patrols may dynamically monitor different segments of a region ; hence, providing a dynamic coverage. If the trespassers randomly move, the best strategy for the border patrols would be to move randomly. However, realistic trespassers typically plan their paths according to the environmental characteristics and past trespassing statistic of the region. They travel on their planned paths rather than travel randomly. A more intelligent trespasser would regularly adapt his traveling path in order to avoid being captured. To overcome the trespassers’ intelligent with adaptive behavior, the border patrols should monitor the region adaptively based on current available knowledge related to the trespassers. Due to a large area and obstacles of the typical border region, precise knowledge of the trespassers’ locations may not be available. In such region, knowledge can be archived through an assumption on known characteristics of the border region and observable footprints derived from discovered traces. This thesis exploits the border patrols’ patrol path planning in two aspects. The first aspect considers the unobservable environment, in which the border patrols has no observation related to the trespassers’ current locations. The border patrols can only rely on the assumption on known characteristics of the region. In this environment, the border patrol’s patrol path planning should be based on the heuristic that is constructed from known characteristics of the region. In addition, the heuristic should be adapted to a dynamic characteristic of the region. The second aspect considers the partially observable environment, in which the trespassers’ locations can be partially observed with uncertainty. In the second aspect, the trespasser is assumed to be adaptive and may not travel fluently based on the known characteristics of the region. A border patrol’s path should be planned adaptively based on a real-time observation of uncertain information. In addition to these two proposed patrol path planning schemes, this thesis contributes necessary models that could be useful for exploiting a border surveillance problem. The empirical experiments were conducted by mean of the simulated agents, which were programmed with the specific behaviors in the simulated border regions. The performances of the proposed patrol planning schemes were evaluated and compared against other patrol path planning schemes in different scenarios. The results show that the proposed patrol path planning schemes outperformed other schemes in term of the trespasser detection rate, where the border region has heterogeneous characteristics and the trespassers are partially observed

Thammasat University. Thammasat University Library

Address: BANGKOK

Email: preserv@tu.ac.th

Name: Komwut Wipusitwarakun

Role: advisor

Created: 2020

Modified: 2022-11-04

Issued: 2022-11-04

วิทยานิพนธ์/Thesis

application/pdf

DOI: 10.14457/TU.the.2020.71

eng

DegreeName: Doctor of Philosophy

Level: Doctoral Degree

Descipline: Engineering and Technology

Grantor: Thammasat University

©copyrights Thammasat University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	9279wichai.pdf	5.3 MB

ใช้เวลา

0.032666 วินาที