Application of Image Recognition and Remote Control to a Service Robot (II)

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

MOST103-2221-E019-048

Translated Name/計畫中文名

結合影像識別及遠端遙控於服務型機器人之應用 (II)

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=8347751

Year

2014

Start date/計畫起

01-08-2014

Expected Completion/計畫迄

31-07-2015

Bugetid/研究經費

817千元

ResearchField/研究領域

電子電機工程
資訊科學--軟體

This research utilizes image recognition, intelligent systems, RFID and WiFi wireless communications to service robots. The purpose of this project is to design service robots that can take elevator, deliver objects, remote control and surveillance. This is the second year of the proposed project. In this year we will continue the development of the first year project and make the robot able to perform patrol and surveillance. In addition, the second year project will develop dual-robot patrols for working on building regular check and backup support each other. Such that the proposed service robots can replace some of security guard’s works and reduce personnel cost. Image recognition is performed by neural network and controller design is based on intelligent system. Image processing is under HSV color space, and pattern recognition is handled by the Adaptive Resonance Theory. Backpropagation through time neural networks are applied to path planning and obstacle avoidance for dual-robot patrols. Blue tooth and WiFi are utilized to robots communications for short-distance mode and long-distance mode. RFID is used for checking room numbers when the visual sensor is unable to identify the numbers due to dim light. In this study, Labview 2010 is utilized for human-machine interface. The wheeled mobile robot is controlled by the calculated angle and distance. This research will develop adaptive neural network that can adopt surrounding changes for image recognition, and provide suitable rules for control strategy. Finally, set up easy structure, fast learning and adequate algorithm for hardware realization, which can reduce production cost. This study uses two wheeled mobile robots as the carriers and integrates ultra sonic sensor, infrared sensor, camera, wireless communication module, and laser range finder to perform auto guidance, path tracking, obstacle avoidance, and remote surveillance. In real time surveillance, Visual studio 2008 will be used to develop interface of image processing, controller, wireless camera, and Zigbee communications. Through image processing technique, object can be identified and the distance can be obtained. This information is then used to control robot arm to reach target button or to drive the wheeled robot to follow the object. Fuzzy system is applied to robot motor controller design. Images of surroundings can be transmitted to a smart phone via WiFi. The smart phone can remote control robot movement and perform surveillance. 本計畫是結合影像識別、智慧型系統、RFID 及 WiFi 無線通訊於服務型機器人之應用，期望設計具備搭乘電梯、物件傳遞、遠端遙控、室內巡邏監視等功能的服務型機器人。本計畫為第二年計畫(103/8-104/7)，延續第一年計畫(102/8-103/7)之搭乘電梯及傳遞公文之功能，繼續研發具備大樓巡守功能之服務型機器人。本期計畫將設計雙機器人巡守隊，從事大樓關燈、巡邏、異象回報、相互支援等工作，以期取代目前保全人員部分的工作，節省人事費的支出。本研究利用智慧型控制及智慧型系統，選擇適合之類神經網路於影像辨識，並以智慧型系統設計控制器。第一年期主要目的是整合 HSV 色彩空間的影像處理、自適應共振理論的圖形識別以及模糊控制系統，應用在輪式人型機器人自動搭乘電梯及傳遞公文。第二年期主要是以 Backpropagation Through Time algorithm (BPTT) 類神經網路，建立雙機器人巡守隊的工作規劃、路徑規劃及避撞控制，並結合藍牙與 WiFi 執行近距離雙人組及遠距離雙人組的互動與聯繫，並提出結合圖形辨識之機器人運動法則，控制機械手臂觸控按鈕及開關。RFID 將被使用於房間辨識，以輔助並解決因光線不足時，機器人不易進行房號視覺辨識的問題。硬體實現將使用 Labview 2010 做為人機介面，利用角度以及距離作為輪式機器人路徑追蹤的依據。本研究要尋找具備適應環境變化能力之類神經網路於影像識別及路徑規劃上，並利用智慧型控制法則訂定控制策略，以期建立結構簡單、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。本研究將應用 2 台輪式人型機器人，結合超音波距離感測模組、紅外線測距感測器、攝影機、無線通訊模組、雷射測距感測器等，進行自動導航、尋跡、避撞、巡邏及監視。在即時室內監視之應用上，本研究將使用 Visual studio 2008 來撰寫人機介面，以 C#語言撰寫影像處理、控制器、無線攝影機和 Zigbee 無線通訊，作為電腦端之間的數據傳遞。經由網路攝影機擷取到的環境影像，利用 HSV 色彩空間轉換，篩選出目標物，再透過影像處理技術與距離演算法，計算機器人與目標物的實際距離，以控制機器人追隨目標物或控制機械手臂觸控牆壁上的按鈕及開關。模糊系統將應用在攝影機的伺服馬達控制與機器人控制器的設計上，機器人可執行監控動態目標物。影像經由 WiFi 無線傳輸傳送至手機上，手機可遠端遙控機器人前進、後退、左右轉，並進行室內巡邏監視。

Keyword(s)

服務型機器人
路徑規劃
圖形識別
遠端遙控
類神經網路
模糊系統
自動導航
Service Robot
Path Planning
Pattern Recognition
Remote Control
Neural Networks
Fuzzy System
Automatic Guidance

DSpace CRIS

Application of Image Recognition and Remote Control to a Service Robot (II)

Details

Description