CNS/EE 248
Sensory Information Processing Laboratory

Spring 1997


Instructors:	Pietro Perona, Demetri Psaltis
	TAs:	Greg Steckman, Mario Munich
	e-mail:	{perona, mariomu}@vision.caltech.edu
		{psaltis, steckman}@sunoptics.caltech.edu
	Web:	http://www.cns.caltech.edu/cns248/.index.html


Robot Navigation

Acknowledgement: This lab was prepared by Greg Steckman and Mario Munich
		Modified by Wenhai Liu and Xiaolin Feng  (Jun. 97)


Abstract

In this project you will construct the algorithms and software necessary to
navigate a robot through the hallways of the subbasement of the Moore
building. This task requires familiarization with the robot hardware and 
camera/framegrabber systems. Throughout the project you will develop
digital image processing routines to be used as the basis for control and
navigation of the robot . 

Weeks 1-2: The robot hardware. Controlling the robot with motion
commands. Using the camera and pan-tilt unit. Capturing images with
the framegrabber. 

  The robot hardware: Read the user and programmer manuals for the
Nomad robot. 
  Robot control: Learn to write software to control the robot
motion. Use the robot simulator to test your routines and familiarize
yourself with the programming interface.
  Pan-Tilt Unit: Read the pan-tilt unit users guide. Write software to
control the motion of the pan-tilt unit.
  Framegrabber: Learn about the operation of the framegrabber and
write software to capture images. 
  Milestone: Demonstrate working robot-simulator software for
controlling the robot movements. Include in your notebook (i) a brief
summary of how the robot works and what your simulator code does, and
(ii) some images that your captured with the robots framegrabber. 

Weeks 3-4: Image processing. Direction initialiation and Kalman filter.
  Read the CNS248 class report by Steckman and Ouyang(1996), Xiaolin
and Wenhai (1997). 
  Develop image processing routines for detecting the molding strip
edges. and determine the direction of hallway, the distances to
left/right wall. 
  Learning the Kalman filter and control algorithm.
  Use the technique described in the paper to get the robot to find
the hallway direction. 
  Milestone: Demonstrate the robots ability to face to hallway and
move to the center line. 

Weeks 5-6: Autonomous navigation and Turning at the end of hallway
  Implement the control algorithm and verify that the robot can walk
straight along the center line of hallway.  Develop an algorithm for
finding the end of hallway and turn to right/left. 
  Milestone: Demonstrate the robots ability to navigate around the hallway.

Weeks 7-8: Improvement of the system.
  Improve the walking algorithm. Develop a algorithm to recognize door
and room number. 
  Make a map of the subbasement and give the robot the capability to
find its current location. 
  Milestone: Implement and demonstrate at least one of these
improvements. 

References

[1] G. Steckman and G. Ouyang, Robot Navigation, Technical
Report CNS/EE248, California Institute of Technology, 1996.
[2] X. Feng and W. Liu , Robot Navigation, Tehnical Report CNS/EE248,
California Institute of Technology,1997. 
[3] E.D. Dickmanns, B. Mysliwetz, and T. Christians, An integrated
spatiotemporal approach to automatic visual guidance of autonomous
vehicles , IEEE Transactions on Systems, Man, and Cybernetics,
20(6) : pp1273-1284, 1990.