Visual Marker Tracking for Robotics

Easily record and analyze the moves of robots, indivitual joints, ...

Similar projects worth following
The idea is to create a smartphone and/or desktop application using Machine Vision tools to track the location of motion tracking markers and then graph, analyze and export the motion data.

This idea came about while working on my Master Thesis on the topic of Jumping Robots. (See link on the left)
We needed a quick and easy tool to capture the location and orientation of a robot platform during its jump, as well as some detailed behavior of its pneumatic actuators.

Fortunately, we had access to a high speed camera. We created a simple Processing sketch to find markers attached to the jumping mechanism and output the data in the console. Later, the movement was graphed in Excel. This process was cumbersome and repetitive. It begged the question: What if we had an app that allowed quick and easy capture of motion profiles using simple machine vision algorithms?

This is a software-heavy project. However, its main intended use is for robotics applications.

Main application screen mockup

Adding a marker to the scene

Visualizing the data, as an x-y-plot or over time

Additional feature ideas:

  • Special "calibration markers" within a known distance from each other for scale calibration
  • Lens distortion feature, initially for commonly used lenses (e.g. iPhone camera)
  • The data can be exported to CSV or other easy to read formats.
  • The analysis can be performed either on a live video stream (e.g. smartphone camera) or on an existing video file.
  • A movie can be exported with markers and data overlaid.

As an engineer working in Machine Vision, I realize marker tracking in the real world is a huge challenge for this kind of algorithm. However, I was surprised at how robust even the crappiest implementation of my convolution matrices was, written with little know-how of the subject. While the emplyed method did not yield sub-mm accuracy, it provided us with a straightforward way to (at least qualitatively) assess our data. And that's good enough a lot of the time!

  • Excerpts from the Master Thesis

    Daniel03/14/2016 at 21:17 0 comments

    This is where the idea originated:

    The aim was to track the expansion of a soft pneumatic actuator to be used in the leg of a jumping robot. Detailed analysis of the accelerating motion was obtained with a high-speed camera.

    It was easy to perform lots of test runs with different parameters (air pressure, payload weight) thanks to the automated motion analysis.

    Using multiple trackers in one scene.

    The jumping robot jumping :-)

View project log

Enjoy this project?



marcusvinciuspigato wrote 03/04/2018 at 12:40 point

Bom dia
Meu nome é Marcus Pigato e sou do Brasil.
Gostaria de verificar a possibilidade de fabricação do item para comercialização aqui.
você está interessado?

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates