Kah Wai@spherebox

A Step Further.

Project Work / Endeavour 2007

Hehe thanks so much to Sian Siew for updating my blog theme. The old one was nice, but the column width a little too small to do any serious blogging. (I don’t want my blogs to be merely 3 lines thick =/ )

Ok… Back to topic. Being a final year student, I’m allocated a project to work on. Along with 2 other teammates, our project will be a year long thing with a final year presentation to the public as well to our supervisors.

Project Name: High Speed Vision System. Project code: RC2 (RC standing for Robocup). Below is a project description by Robert Li, one of my group members:

“One of the most important aspects of a robot is its sensing/vision system.

This project is to realise sophisticated object recognition algorithm on hardware which is designed to be used on robotic platform but will also be compatible with a board range of industry applications. Highly accurate ball tracking and moving target recognition at high speeds is the ultimate goal of this project.

By using a high resolution video camera and a specially designed mirror, robots are able to “see” a 360 degree image (essentially a top down view of a defined area). A Field Programmable Gate Array (FPGA) is used as a primary mean of speeding up the process of object detection. It’s extraordinary concurrent processing feature allows processing different coloured objects simultaneously and pipeline processing each pixel as it comes into system. External Synchronous RAMs are used to support large volume of data while maintaining high speeds.”

Since we were needed to submit our own project descriptions, I believe the text above neatly summarises our project aims in a nutshell. That being said, the complexity of it is enormous as we are going to integrate the algorithm into hardware (the algorithm being realised last year by a group of students, Ni Ma and Nathan Williams. Their project website: http://fpga-vision.sourceforge.net/

Current issues that we are addressing:

  • Bus size to interface with the processor board. (which is managed by Boon Chong Khor, Nicholas Seng Tatt Yeoh and Shubham)
  • Building our very own development PCB board since the RAM chips used in the stock dev board is DDR-RAM and we’ll be using SRAM.
  • Timing issues with processing frames ‘as they come’ (hence high speed) as well as application consideration (the speed which is required by the application)

The vision system that we will develop will ultimately be used in the robots that are representing my university in future Robocup tournaments. As I go along with the project, I will post some pictures of our progress as well as the actual robot itself.