Vision-Based Control of Mobile Robots

	D. Burschka and G. Hager
	Computational Interaction
	and
	Robotics Laboratory (CIRL)
	Johns Hopkins University

Outline

	Introduction
			Navigation Strategies
	Principles of the
	Vision-Based Control (VBC)
	Tracking in Monocular Images
	Generation of the Control Signals
	Results
	Conclusions

Navigation Strategies

How does Vision-Based Robot Control work?

XVision as Tracking Tool

How does Pattern Tracking work?

Virtual Image Plane

Extension of the field of view

Independence from the Physical Sensor Properties

Image Jacobian

Estimation of yi

Generation of Control Signals

Simulation Results

Example Application

Conclusions and Future Work:

	System for robust navigation based on monucular sensor perception is presented
	Signals are generated directly from the image
	Extension to omnidirectional sensor
	No exact calibration is necessary
	Integration of automatical landmark selection (Swain, Kriegman)

Additional Information:

	Web:
	http://www.cs.jhu.edu/~burschka
	http://www.cs.jhu.edu/CIRL