The Office Marathon: Robust Navigation in an Indoor Office Environment Marder-Eppstein, Eitan, Berger Eric, Foote Tully, Gerkey Brian P. , and Konolige Kurt International Conference on Robotics and Automation, 2010 Zhou Xin 2011. 5. 13
The authors come from… • Willow Garage, Inc. – http: //www. willowgarage. com • working on – – – – Perception Motion planning Navigation Hardware integration Simulation Visualizers And more…
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Problem Description – Reachable goal – Unmodified office-like environment – Autonomous navigation – Avoiding obstacles • Robot can’t step over – 3 D Obstacles • Can change location
Problem Description • Rounding a blind corner • Cluttered environment -Unknown space -Various obstacles
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Key Challenge • Detecting and Avoiding • The non-trivial 3 D structure obstacles
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Platform • 8 -wheeled omni-directional base • 2 Laser scanner • ROS(Robot Operation System)
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Approach • A. Sensor Processing Pipeline – Sensor is imperfect – Accuracy: 1. 5 cm – Random SAmple Consensus (RANSAC[1] ) algorithm
Approach • Core—Voxel Grid – Each cell in Grid • Occupied • Free • Unknown – Two dimensional array of 32 -bit – Operations • Marking • Clearing(Bresenham[2])
Approach -- Policy • Entrapment – In-place rotation – Clear obstacles • Recharging – Email a plug-in and unplug request.
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Experimental Results Table 1. Results for the experiment environments both in simulation and the real world
Outline • • • Problem Description Key Challenge Platform Approach Experimental Results Conclusion
Conclusion • A robust navigation system is proposed • Use efficient Voxel-based 3 D mapping • Limitation: – Can not fully get rid of FALSE NEGATIVES. – Can’t detecting when it is stuck.
![REFERENCES • • [1] M. Fischler and R. Bolles, “Random Sample Consensus: A Paradigm](https://present5.com/presentation/920ab76e3cff7cf82b832b1aa8512795/image-18.jpg "REFERENCES • • [1] M. Fischler and R. Bolles, “Random Sample Consensus: A Paradigm")
REFERENCES • • [1] M. Fischler and R. Bolles, “Random Sample Consensus: A Paradigm for Model Fitting with Application to Image Analysis and Automated Cartography, ” Communications of the ACM, vol. 24, pp. 381– 395, 1981. [2] J. Bresenham, “Algorithm for computer control of a digital plotter, ” IBM Systems J. , vol. 4, no. 1, pp. 25– 30, 1965.
Скачать презентацию The Office Marathon Robust Navigation in an Indoor
920ab76e3cff7cf82b832b1aa8512795.ppt