Related Experiment Videos
Tracking a Non-Cooperative Target Using Real-Time Stereovision-Based Control: An Experimental Study.
1Department of Aerospace Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel. tomershtark@gmail.com.
Sensors (Basel, Switzerland)
|April 1, 2017
Summary
This study presents a real-time stereovision system for tracking unknown targets. It enhances target tracking accuracy using the unscented Kalman filter and total least squares for improved navigation and control.
Area of Science:
- Robotics and Computer Vision
- Sensor Fusion and Estimation Theory
Background:
- Tracking non-cooperative targets in unknown environments is difficult.
- Stereovision offers passive, wide-area scanning for object localization.
Purpose of the Study:
- Develop and evaluate real-time non-cooperative target tracking using only stereovision.
- Improve accuracy in static and dynamic stereovision triangulation and line-of-sight control.
Main Methods:
- Implemented a computer-vision algorithm for feature detection and matching.
- Designed and compared three filters for position and velocity estimation.
- Utilized a line-of-sight control algorithm to maintain target visibility.
Main Results:
- The unscented Kalman filter combined with total least squares significantly improved stereovision-based tracking.
- Demonstrated benefits through experimental and numerical evaluations.
- Showcased accurate solutions for static and dynamic stereovision triangulation.
Conclusions:
- Stereovision-based tracking is effective for non-cooperative targets.
- The unscented Kalman filter and total least squares provide a robust and accurate solution.
- Findings contribute to general methods for stereovision triangulation and control.