Application of Image Processing and Vi Sual Navigation to Indoor Security Robot

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

MOST109-2221-E019-058

Translated Name/計畫中文名

影像處理及視覺導引於室內保全機器人之應用

Project Coordinator/計畫主持人

Jih-Gau Juang

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Communications, Navigation and Control Engineering

Website

https://www.grb.gov.tw/search/planDetail?id=13544351

Year

2020

Start date/計畫起

01-08-2020

Expected Completion/計畫迄

31-07-2021

Bugetid/研究經費

802千元

ResearchField/研究領域

電子電機工程

Based on the projects that have been completed in the past, this study applies image identification methods to integrate appropriate sensors and neural networks and fuzzy systems to preserve the design and application of robot path guidance, obstacle avoidance and control laws. The main purpose is to find intelligent rules that have the ability to adapt to environmental change, and to establish appropriate calculations that are simple in structure and fast to learn, so that costs can be minimized when hardware is implemented. Previous project has completed the function of automatic taking elevator and documents delivering by the proposed robot system, but this project will use visual servo control to autonomous navigation and obstacle avoidance control with dual-lens cameras, which require different ways of planning. The robot needs to carry on the indoor path planning and the guidance through the feature matching algorithms. The depth camera provides environment distance for obstacle avoidance. In this study, a simple wheeled mobile robot system, combined with an image sensing module, is used for automatic navigation, collision avoidance, patrol, open door, turn off the light switch and other indoor services. The robot arm is applied to open door and turn off light by fuzzy system and visual servo control. In the first phase of this project, conventional image processing methods will be applied. In the second phase, deep learning neural network will be used in object identification. The robot can be controlled by the identified images which bring out the corresponding action that the robot needs to follow. The robot then follows the commands and performs related actions, the indoor service mission can then be accomplished. 本研究是以過去已完成的計畫為基礎，利用影像辨識方法，整合適當之影像感測器與類神經網路及模糊系統，於保全機器人路徑導引、避障及控制法則之設計與應用。最主要的目的是要尋找具備適應環境變化能力之智慧型法則，建立結構簡單、學習快速的適當演算方式，以便在硬體實現時，可使得成本降至最低。前期計畫已完成輪式機器人自動搭乘電梯及傳遞公文的功能，但本計畫將以雙鏡頭攝影機進行自主導航與避障控制，需要不同方式的規劃。本研究應用簡便型輪式機器人，結合雙攝影機進行自動導航、避撞、巡邏、開門、關燈等室內服務型機器人之應用。機械手臂開門及關燈的控制，是結合模糊系統與視覺伺服控制。本計畫第一階段以傳統影像處理方法進行特徵點匹配，第二階段則利用單一簡化型深度學習神經網路，學習整個室內走廊及教室的特徵點圖像，將這些圖像標記成相對應的行動序列，當機器人辨識出圖像標記時，同時亦獲得該圖像所給予的接下來之行動指令，或轉彎或直行或開門或關燈，機器人即可依序執行預先設定的任務，如檢視走廊及關燈等，以達成室內巡守之目的。

Keyword(s)

視覺伺服控制
避障控制
路徑導引
影像處理
模糊理論
自動導航
輪式人型機器人
visual servo control
collision avoidance control
path guidance
image process
automatic guidance
wheeled mobile robot

DSpace CRIS

Application of Image Processing and Vi Sual Navigation to Indoor Security Robot

Details

Description