FastTrack:

1
FastTrack A High Frame Rate Stereovision
Tracking System
Peter Chun Valeri Kirischian Vadim Geurkov Lev
Kirischian Dept. of Electrical Computer
Engineering Ryerson University
Michael Belshaw Michael Greenspan Dept. of
Electrical Computer Engineering Queens
University
Siraj W. James MacLean Dept. of Electrical
Computer Engineering University of Toronto
2
Objective
Develop a stereovision system capable of
extracting range data at gt 200 fps and tracking
objects
- state of the art for stereovision extraction
30 fps
Motivation Autonomous Satellite Grasping
Benefits Increasing the speed of tracking
will 1. Allow us to track faster moving
objects 2. Allow us to track slower moving
objects more reliably
3. High frame rate stereovision and tracking
will enable other applications
3
Fast Track Satellite Grasping

• Automatic image recognition and docking system
• Ultra-fast (200 fps) stereo video acquisition
• Real-Time Video-Stream processing on the FPGA
  based platform

4
Propelling Collaboration
2nd generation SVAR project
Object Tracking Algorithms
200 fps Stereo Camera
FPGA Platform
Fast Stereo-Vision Algorithms
5
System Overview
200 frame/sec. Stereo-Video Camera
Interface Module
Video-Output Module
FPGA-based Video-processing Platform
FPGA Xilinx Virtex-II Pro
PCI-bus Interface
Object in the are of Stereo-vision
Host Computer
6
FPGA Platform
7
Stereovision
8
Object Tracking
Two methods ICP and BHT - focus on ICP - more
standard - less complex to implement on FPGA
9
Project Timeline