Visual Servoing for Flexible Robotic Assembly Tasks, 10-9410

Printer Friendly Version

Principal Investigators
Brent M. Nowak
Michael P. Rigney
Jeremy K. Zoss
Nicholas R. Gans
Ian R. Mienzen

Inclusive Dates: 07/01/03 - Current

Background - The objective of this project is to develop a visual servoing capability to enhance the performance of robotic assembly tasks. Visual servoing is defined as the use of machine vision to provide closed-loop control of a robot end-effector or mobile platform. Industrial robots typically operate "blind," having no sensory input regarding the positions of objects within their workspace. Subsequently, a large part of the cost for robotic installations is fixtures, tooling, and material-handling mechanisms. Visual servoing can provide both increased accuracy to a robotic manipulator and allow the robot to function in a less constrained environment.

Approach - The objectives of the proposed program are to:

  • Develop a visual servoing capability and verify that the technique is a valid solution in industrial applications using open-architecture controllers.

  • Demonstrate that visual servoing can provide performance appropriate for complex control of manufacturing processes using the current generation of robot controllers and machine vision systems.

Accomplishments - A visual servoing test bed, which includes an industrial robot with an open-architecture controller and a PC-based vision system has been integrated and configured. Images are captured by a CMOS camera providing full-image (640 x 480 pixel) frame rates up to 105 frames per second. This frame rate offers the potential for shorter servo loop timing than could be achieved from a conventional 30 Hz camera. An adjustable camera-mounting bracket for a robot end-effector was designed and installed (Figure 1). A calibration procedure, target, and software have been developed. Software demonstrating the execution of a simple visual servoing task has been developed and tested (Figure 2).

Figure 1. An adjustable camera mount was designed for the robot end effector. The camera mount is compatible with task specific grippers or end effector tooling (not shown).

Figure 2. The visual servoing application user interface provides high-level application control. A sample image from a simple visual servoing test application is shown with machine vision results superimposed.

2003 Program Home