Visual Servoing (VS) has been researched for over thirty years, but real-world adoption of VS has been slow. Challenges include camera calibration, lack of, or difficulty integrating reliable real-time visual trackers and a lack of simple control interfaces to which robots can be connected. In this work we present a minimalistic framework for Uncalibrated Visual Servoing (UVS) with the hope that we can alleviate some of the issues that are holding back adoption. We focus on enabling users to create simple control interfaces that easily interact with tracking systems to produce a complete system. Our Visual Servoing library, ROS-UVS, has now been used with several kinds of robots for a variety of tasks such as pick and place with a WAM manipulator and quadrotor assisted flight control. Our implementation, developed within the ROS framework, has proven to be flexible, robust and easy to use and integrate across multiple robots.

Motivation

• Minimalistic: only one external dependency. Eigen for linear algebra
• Focus on the visual servoing. Do one thing and do it well motto
• Avoid including trackers and all their complications like software dependencies.
• Based on a single header file when used as a stand alone library.
• Conects directly and easily to robots. We have used it across many applications. Having few dependencies makes it easier to adapt across different projects.
• The library is designed with portability, usability and performance in mind. Ease of use is also crucial.

Main Features

• No a-priori models required
• Uncalibrated Visual Servoing Implementation
• General UVS library adaptable to any robot
• Ability to run simulations in Matlab
• Ability to run as a stand alone library
• No dependencies outside of a base ROS installation
• Support for any number of cameras and error inputs

Design Strategy

Resources

• ROS-UVS
• Eigen