Mobile robot using different senses Motivation Senses for Robots - PowerPoint PPT Presentation

Mobile robot using different senses Motivation Senses for Robots ISOEN 2002 � Sight (Cameras) � Hearing (Microphones) M. Wandel*, A. Lilienthal + , A. Zell + , U. Weimar* � Touch (Tactile sensors) *ipc, AG Weimar � Smell (Gas sensor system) www.ipc.uni-tuebingen.de/weimar/ � Taste (Electronic tongue) + WSI-RA, AG Zell www-ra.informatik.uni-tuebingen.de Extending the possibilities � Electronic Watchman can oversee chemical stocks and detect leakages 1 University of Tübingen Institute of Physical Chemistry 3 University of Tübingen Institute of Physical Chemistry Overview Robots with Gas Sensors: Applications Potential Applications Biomimetic • Motivation • Trail following • Path Finder � State of the art � Trail defined life time • Virtual Umbilical • Experimental set-up • Marking • Repellent Marker � Mobile robot � Mark of Cleaned Floor � Sensor system • Source Detection • Experiments � Direction finding � 1-Dimensional experiments in a corridor � Rotating Robot � 2-Dimensional experiments in large room Short Distance (Duckett et al.) • Findings � Three dimensional Odour Compass – Defined air stream (Nakamoto et al.) • Conclusions � Detection of odour plumes • Outlook � Plume tracking (Russel et al.) – Defined air stream needed 2 University of Tübingen Institute of Physical Chemistry 4 University of Tübingen Institute of Physical Chemistry

Robots with Gas Sensors: Applications Experimental: Sensor System Source Detection Potential Applications Biomimetic VOCmeter Vario • Direction finding • Trail following • Path Finder • Commercially available � Rotating Robot � Trail defined life time • Virtual Umbilical Short Distance (Duckett et al.) • Lightweight, small • Marking � Three dimensional Odour Compass • Repellent Marker • Low power consumption � Mark of Cleaned Floor � Defined air stream (Nakamoto et al.) (24V DC supply possible) • Source Detection • Detection of odour plumes • RS-232 interface � Direction finding � Plume tracking (Russel et al.) � Rotating Robot � Defined air stream needed • Various sensors types Short Distance (Duckett et al.) • Convection / Diffusion profile available (MOX, QMB, U/I) � Three dimensional Odour Compass � Map Building – Defined air stream (Nakamoto et al.) • Up to 8 sensors, connected � No Defined Airstream needed � Detection of odour plumes with thin flexible cables � Plume tracking (Russel et al.) – Defined air stream needed 5 University of Tübingen Institute of Physical Chemistry 7 University of Tübingen Institute of Physical Chemistry Experimental: Mobile Robot “Arthur” 1-Dimensional Location Locations Mobile Robot “ATRV-Jr” • Skid Steering � No ventilation! • Standard PC inside � No personnel traffic • Ultrasonic sensors • Wireless Ethernet Experiments Additional Sensors • Analyte used: Ethanol • Laser range finder • Recordings • Stereoscopic camera � odometry, sensor system, laser range finder, System � temperature, humidity (offline) • Gas sensor system mounted inside the robot, 60cm behind front window 6 University of Tübingen Institute of Physical Chemistry 8 University of Tübingen Institute of Physical Chemistry

Results: 1-Dimensional Location Results: 1-Dimensional Location 14 14 0 0 12 12 10 10 -5 -5 8 8 ∆ n [ppm] ∆ n [ppm] -10 -10 s [m] s [m] 6 6 4 -15 4 -15 2 2 -20 -20 0 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 t [s] t [s] 9 University of Tübingen Institute of Physical Chemistry 11 University of Tübingen Institute of Physical Chemistry Results: 1-Dimensional Location Determination of Source Position 14 14 0 10 4 0 12 12 10 -5 10 5 -5 3 8 8 ∆ c [ppm] ∆ c [ppm] ∆ n [ppm] -10 -10 0 s [m] 6 6 s [m] s [m] 2 4 -15 -15 -5 4 2 2 -20 1 -20 -10 0 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 2500 2750 3000 3250 3500 3000 3100 3200 3300 t [s] t [s] t [s] 10 University of Tübingen Institute of Physical Chemistry 12 University of Tübingen Institute of Physical Chemistry

2-Dimensional Location Comparison Summer - Winter • Unventilated Summer Winter • One side of the room is a window front • The automatic source was placed in the middle of the room • The robot’s path describes a rectangular helix ] signal [a.u.] . u . a [ l a n g i s 0 0 2 2 4 4 5 5 6 6 y 4 4 8 3 8 3 2 2 10 10 1 1 12 0 12 0 x 13 University of Tübingen Institute of Physical Chemistry 15 University of Tübingen Institute of Physical Chemistry Results 2-Dimensional Location Peaks / Position 4,0 3,5 signal [a.u.] 3,0 Y [m] 2,5 2,0 1,5 Summer 0 1,0 Winter 2 4 5 6 4 8 3 2 0 2 4 6 8 10 12 10 1 12 0 X [m] 14 University of Tübingen Institute of Physical Chemistry 16 University of Tübingen Institute of Physical Chemistry

Time of Occurrence Conclusion Signal (baseline) • In a 1-dimensional environment the position Signal (peak) Online computed baseline of the source can be estimated within a range Offline fit of 1m • In 1- and 2-dimensional environment the signal / a.u. time of occurrence of the leak can be computed with good precision • The measured concentration profile requires time consuming search strategies e.g. map building, this is possible because of the stable concentration profile 0 1000 2000 3000 4000 t open, predicted 0 1000 2000 3000 4000 t open, real time / s t [s] 17 University of Tübingen Institute of Physical Chemistry 19 University of Tübingen Institute of Physical Chemistry Findings Outlook • Signal nearly independent of stick position • Implementation of a feedback from the sensor signal evaluation to the driving • Simple mounting of the sensor, without fan, programme without pumping unit, sufficient • Larger rooms • Meaningful measurements only during movement • Classification of the analyte • Even small leaks can be detected � Using a sensor system with different sensors including QMBs and pattern recognition software • Testing in real world environment 18 University of Tübingen Institute of Physical Chemistry 20 University of Tübingen Institute of Physical Chemistry

Acknowledgements Landesforschungsschwerpunktsprojekt Baden-Württemberg Sensing organs for mobile robots Project partners: • Wilhelm-Schickard-Institut für Informatik, AG Zell (coordinator) • Zoologisches Institut, AG Schnitzler • Institut für Physikalische und Theoretische Chemie, AG Weimar 21 University of Tübingen Institute of Physical Chemistry Thank You 22 University of Tübingen Institute of Physical Chemistry