Experiences Using Experiences Using Gas Sensors as Sensors on an - PowerPoint PPT Presentation

Experiences Using Experiences Using Gas Sensors as Sensors on an Autonomous on an Autonomous Mobile Robot obile Robot Achim Lilienthal Achim Lilienthal University of Tübingen University of Tübingen WSI Computer Science Dept. WSI Computer Science Dept. http://www-ra.informatik.uni-tuebingen.de http://www-ra.informatik.uni-tuebingen.de

Contents � 1. Defining the Goal: An Electronic Watchman � 2. Hardware Setup � electronic nose � mobile robot � 3. Previous Experiments (1D) � 4. Experimental Setup (2D) � 5. Results (2D) � 6. Conclusion & Outlook Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 2

1. Defining The Goal: An Electronic Watchman � Required Ability � detection of gases � Desired Abilities � localization of the gas source � identification of the odour � Environment unmodified indoor environment � Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 3

2. Hardware Setup, Electronic Nose � VOCmeter Vario � commercially available � lightweight, small � 24V DC supply possible � low power consumption � operates up to 8 sensors � gathers readings with 4 Hz � RS-232 interface Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 4

2. Hardware Setup, Electronic Nose � Sensor Sticks � thin, flexible cable � small, easy to mount � MOX, QMB available Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 5

2. Hardware Setup, ARTHUR � Based on "ATRV-Jr" (RWI) � skid steering � 2x Pentium II, 333 MHz � wireless LAN (BreezeCOM) � sonar sensors 6 5 7 8 � Electronic Nose 4 � sensors: MOX – at an outstanding rotatable bar 3 – at fixed positions � Additional Sensors 2 75 cm � laser scanner (SICK) 1 � CCD cameras Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 6

3. Previous Experiments (1D), Setup � Experimental Conditions � no / weak ventilation � no / few people passing by � Odour Source � ethanol, aceton � placed at the end / in the middle of the corridor � different intensities: 130 cm 2 , 60 cm 2 , 20 cm 2 � Driving Modes � stop-and-go � constant velocity � Gas Sensors � mounted on the stiff extension Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 7

3. Previous Experiments (1D), Results Ethanol Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 8

3. Previous Experiments (1D), Results � Detection � low intensities � distance: several meters � unventilated or weakly ventilated rooms � weak disruptive elements possible � Localization � seems to be possible Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 9

3. Previous Experiments (1D), Results Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 10

3. Previous Experiments (1D), Results � Driving Mode � constant speed, not too slow � stop-measure-and-go strategy not suitable Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 11

3. Previous Experiments (1D), Results Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 12

4. Experimental Setup (2D) � Experimental Conditions � unventilated room, no people passing by � jar in the middle of the room � robot moved along a rectangular spiral 12.90 m 5.20 m 7.10 m Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 13

5. Experimental Setup (2D) Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 14

5. Results (2D) Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 15

6. Conclusion � Setup � signal nearly independent of stick position � meaningful results only during movement � no improvement could be detected using either a – pumped cell – pc fan � Detection and Localisation in a '1D Environment' � even small sources (about 20 cm 2 ) � distance of about 5 m � possible after hours � Detection and Localisation in a '2D Environment' � distinctive peaks along pathways level with the odour source � measured distribution not centered at the location of the source Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 16

6. Outlook � Implementing Intelligent Search/Navigation Strategies � dealing with a partially unknown airflow situation � Investigate the "Moving Effect" � further testing with the pumped cell (trunk) � Identification � using sensor array: MOX and QMB � Environment � less artificial indoor conditions � outdoor conditions Achim Lilienthal, University of Tübingen, WSI Computer Science Dept. 17

Experiences Using Gas Sensors Experiences Using as Sensors on an Autonomous on an Autonomous Mobile Robot obile Robot Achim Lilienthal, Andreas Zell Achim Lilienthal, Andreas Zell University of Tübingen University of Tübingen WSI of Computer Science WSI of Computer Science http://www-ra.informatik.uni-tuebingen.de http://www-ra.informatik.uni-tuebingen.de Michael Wandel, Udo Weimar Michael Wandel, Udo Weimar University of Tübingen University of Tübingen Institute of Physical Chemistry Institute of Physical Chemistry http://www.ipc.uni-tuebingen.de/weimar http://www.ipc.uni-tuebingen.de/weimar

Thank you! Thank you!