Navigation and obstacle avoidance system for sidewalk mobile

Navigation and obstacle avoidance system for side-walk mobile robot Pakorn Udsatid 5370454421 Advisor: Nattee Niparnan, Ph. D. 1

Outline Introduction Related works Problem Etc. 2

Robots – traditional http: //malct 32. blogspot. com/2011/06/developed-economy-generates-more-jobs. html 3

Robots – brand new Office Tour guide Supermarket Autonomous wheelchair [H. Asoh et al. , 01] [A. Clodic et al. , 06] [S. Nishimura et al. , 07; M. Goller et al. , 10] [Y. Kobayashi et al. , 09] 4

Characteristics cooperate with people; move Nice! make people feel comfortable http: //www. theage. com. au/news/technology/the-future-ishere/2006/11/05/1162661536462. html http: //5 magazine. wordpress. com/2011/02/18/roila-the-rob -interaction-language/ 5

Person following vs side-walk [R. Gockley et al. , 07] http: //www. superstock. com/stock-photosimages/1889 R-23044 6

Person following vs side-walk http: //www. fastcompany. com/1587924/robots-codyhttp: //www. flickr. com/photos/21617980@N 05/209928566 manipulator-hands-balancing-georgia-tech-healthcare-home 8 -robot 7

Person following vs side-walk 86 % of people prefer robot to move side by side friendly easy to talk frank Robotic Wheelchair Based on Observations of People Using Integrated Sensors [Y. Kobayashi et al. , 09] 8

Person following vs side-walk http: //www. youtube. com/watch? v=3 x. Xhn. Ju. Pl. V 4 Robotic Wheelchair Based on Observations of People Using Integrated Sensors [Y. Kobayashi et al. , 09] 9

Goal To develop a robot system which is able to move side-by-side person 10

Our problem Develop system for robot including walk beside avoid person obstacles Person tracking Navigation 11

System Overview Sensor Robot measurement data Person tracking Person position Robot control Obstacle detection Obstacles position Navigation 12

Using camera [C. Schlegel et al. , 98] [T. Yoshimi et al. , 06] [W. Zajdel et al. , 05] 13

Using LRF [J. H. Lee et al. , 06] [D. F. Glas et al. , 07] 14

Sensors sensor Laser Range Finder camera data type range price expensive accuracy High image vary moderate http: //robotics. umass. edu/? n=Main. Robotics. Lab. Facilities http: //www. lopburi 1. go. th/school/banchee/input_page_type. html 15

Kinect http: //www. arip. co. th/news. php? id=411497 16

Kinect Advantage depth image affordable price Disadvantage narrow field of view not robust to sunlight 17

System Overview Sensor Robot measurement data Person tracking Person position Robot control Obstacle detection Obstacles position Navigation 18

Person tracking [A. Carballo et al. , 08] 19

System Overview Sensor Robot measurement data Person tracking Person position Robot control Obstacle detection Obstacles position Navigation 20

Navigation 21

Navigation [Cao Qixin et al. , 06] [S. S. Ge and Y. J. Cui, 02] 22

Objective To develop a robot system which is able to move side-by-side person and avoid obstacles and uses kinect as sensor 23

Proposed method 24

Set up Person 25

Challenge Kinect as sensor field of view, sunlight which person model? Move beside person changing direction obstacles Differential drive robot 26

System Overview Sensor Robot measurement data Person tracking Person position Robot control Obstacle detection Obstacles position Navigation 27

Person tracking use data at 2 levels 28

Person tracking Kalman filter Predict Observation Update 29

System Overview Sensor Robot measurement data Person tracking Person position Robot control Obstacle detection Obstacles position Navigation 30

Navigation Virtual force field 31

Evaluation Tracking Measure robot’s path and compare with ground truth data Avoiding obstacles Avoid static objects Person’s feeling 32

Scope Deliver robot system Use kinect as only kind of sensor Conduct experiment in indoor No clutter in workspace 33

Work plan Study the works in the related fields Develop a new method Test the system Evaluate and summarize the result Conclude the research, Write a thesis 34

Thank you. 35