Embedded Vision Architecture

This project investigates high-speed vision systems for visual servo control. It studies the effects of measurement error and measurement delay of visual feedback on control performance.

PhD Candidate: Zhenyu Ye
Graduated PhD student: dr. Roel S. Pieters
Researcher: Mark Wijtvliet
Supervisors: prof. dr. Henk Nijmeijer and Pieter Jonker (Delft Biorobotics Lab)
Project Funding: NWO
Project Period: September 2009 - February 2014

Visual sensing is widely used in mechatronic and robotic systems. It can observe certain on-line dynamics of the system and its environment, which is otherwise difficult or even impossible to observe using other sensors. Despite decades of research, it remains challenging to integrate visual sensing into mechatronic and robotic systems. Several open problems are tackled in this project:

  • The trade-off between accuracy, sample rate, and delay of visual sensing under computational constraints.
  • Cross-domain modeling of precision visual servo systems.
  • Exploration of design choices with regard to overall performance and cost.

This project is performed in cooperation with other partners. More information is available at the homepage of the project.