Software Design For IGVC Robot

1
Software Design For IGVC Robot

• By Santosh B. Nair
• Course MCS 6514 at Lawrence Tech University
• Guide Dr. Chanjin Chung
• Date 05/02/2003

2
Main Functionalities

• The software for IGVC robot is implemented on
  VC using MFC. It consists for these key
  components
• Vision processing
• IR sensor processing
• Motor processing
• Decision processing

3
Class Definition

• CogitoBot MFC Application Class
• CogitoBrain Window Class
• CogitoVision Window Class
• CogitoSensor Window Class
• CogitoMotion Generic Class
• Comport Generic Class
• IR Distance Sensor Generic Class

4
CogitoBot App Class

• This creates the Main window on startup with
  various Menu options.
• The Menu options include
• File
• Start Cogito (Initialize)
• Stop Cogito (Clean up)
• Exit
• Autonomous
• Go (Start Vehicle)
• Stop (Stop Vehicle)
• Navigation (not implemented)
• Follow-the-Leader (not implemented)
• This class instantiates all the other classes,
  when File-Start Menu option is selected

5
CogitoBrain Class

• This class starts 2 threads
• For Camera/image capture pre-process
• For IR Data capture
• Each thread then invokes the Vision or Sensor
  processing routine of this class
• When a Stop command is received (from the
  Autonomous-Stop option), the Robot status is
  set from Active to Stop. This terminates the
  thread and stops the vehicle.
• Vision processing Routine gets an array of 4X4
  (created by Vision class) which contains the
  white pixel count in each segmented region of the
  image
• Based on the white pixel presence in various
  segments, the Robot sets various states and
  actions.
• The actions of the Robot are based on white pixel
  counts in each image segment, and the previous
  states and actions
• This logic was introduced to reduce the veering
  of the vehicle and have a smooth movement.

6
CogitoVision, Sensor, Motion Classes

• Vision class captures two images (from two
  cameras) and merges them into one single image of
  almost twice the width. The height of the image
  is clipped from top and bottom.
• Creates a 4X4 array to store white pixel count,
  representing 16 cells of the segmented image
• Other adjustments like threshold can be done
  dynamically during runtime.
• Uses Intels vision libraries (OpenCV and IPL)
• Sensor class collects the IR Data using the IR
  interface class (provided by Maurice). Data is
  also displayed on the window
• Motor class allows to send commands to control
  motor like Forward, Reverse, Left, Right
  (provided by Andrey). It uses the Parallel port
  interface dll to send the commands to the port.

7
Acknowledgements

• Dr. Chanjin Chung, for the project guidance
• Andrey and Maurice for the interface classes
• All team members for help during the project