人工免疫系統之行動機器人路徑規劃應用

Artificial Immune System Application on Mobile Robot Path Planning

沈玉昌(Yu-Chang Shen)

指導教授 ： 徐元寶

國立虎尾科技大學/電機資訊學院/光電與材料科技研究所/碩士(2011年)

https://doi.org/10.6827/NFU.2011.00146

摘要

本文主要研究如何利用人工免疫系統(Artificial Immune System)，解決機器人在結構化空間中的避障與路徑規劃問題。如同生物免疫系統，生物會以不同的抗原呈現細胞(Antigen Presenting Cell)試探外來的抗原(Antigen)，以獲得足夠的資訊並產生對應的抗體(Antibody)。AIS系統將受控體所面對的環境狀況視為一種抗原，利用感測器蒐集的資訊組成抗原呈現細胞，提供詳細的抗原資訊供AIS系統利用。免疫系統內每一種不同的抗體對同一種抗原具有不同的吸引力及排斥力，經過推算後，得到針對抗原(當前環境)的最佳抗體，再將對應的抗體轉為適當的控制量，以控制受控體，最後達成控制目標。
本研究利用差動輪機器人作為受控體，機器人所置身的環境情境作為抗原，而抗原呈現細胞則分別為，(1)雷射測距儀所量測之機器人與周圍障礙物的距離；(2)利用室內定位系統推算出機器人頭部與目的地的夾角。其中機器人與障礙物的距離視為一種排斥力，機器人頭部與目的地的夾角則視為一種吸引力，利用吸引力與排斥力相互影響的結果，選擇出最利於目前狀況的最佳移動方向(抗體)，作為機器人的移動依據。移動完畢後重複先前的步驟，最終使機器人成功避開環境中的障礙物，並到達設定的目的地。軟體模擬及實驗結果顯示本系統能成功完成多種障礙環境的避障並抵達目的地，證實本方法的可行性。

關鍵字

人工免疫系統 ； 避障 ； 路徑規劃

並列摘要

This thesis focuses on how to use the Artificial Immune System (AIS) to solve the obstacle-avoidance and path-planning problem of a robot in structural spaces. Like the immune system of organisms using different Antigen Presenting Cells (APCs) to test an Antigen in order to obtain enough information to produce a corresponding Antibody, an AIS treats the environment which the plant are encountering being an Antigen and employs sensors to collect environmental information to form APCs for further use. Every different kind of Antibody has different attractive force and repulsive force for the same kind of Antigen in the AIS. After calculation, the optimum Antibody for an Antigen (current environment) is obtained and transformed to appropriate control value to control the plant. Finally, by repeating the process, the control goal is achieved.
This research utilizes a differential wheeled mobile robot as the controlled plant and the environmental situation where the robot is situated in as the Antigen. Antigen Presenting Cells are being: (1) Laser Scanner measured distance between the surrounding obstacles and the robot, (2) the angle between the robot's heading direction and the direction pointing to the destination. The distances of (1) are considered as repulsive forces and the angle of (2) is considered as an attractive force. The interaction result of the attractive force and repulsive force of the current situation is employed to select the best movement for the robot to take. Repeating previous steps, finally, the robot will successfully avoid the obstacles and reach the destination.
Simulation and experiment results show that the system can accomplish obstacles avoidance and reach the destination in variety of environments, proving that the feasibility of the method that this research provides.

並列關鍵字

Artificial immune system ； obstacle avoidance ； path planning

參考文獻

[5] 張政元、羅展雄，”清潔機器人之地形辨識與路徑規劃”，2010電腦視覺、影像處理與資訊技術研討會，2010。

[4] 林建州、蕭博元、林俊立，”以位能場模型為基礎之群組機器人移動規劃”，第九屆離島資訊技術與應用研討會，2010。

[6] N. K. Jerne, “The immune system,” Scientific American, Vo1.229, No.1, pp.52-60, 1973.

[7] J. D. Farmer, N. H. Packard, A. S. Perelson, “The immune system adaptation and machine learning,” Physica 22-D, 1986.

[8] A. Ishiguro, Y. Watanabe, Y. Uchikawa, “An immunological approach to dynamic behavior control for autonomous mobile robots,” IEEE/RSJ International Conference on Intelligent Robots and Systems, Pittsburg, PA, pp.495-500, 1995.