Clarke, Tim 
THE CONTROL OF A ROBOT END-EFFECTOR USING PHOTOGRAMMETRY 
Dr. T. Clarke & Dr. X. Wang 
Optical Metrology Centre, City University, 
Northampton Square, London, EC1V 0HB, UK 
t.a.clarke(@city.ac.uk, x.wang@city.ac.uk 
KEY WORDS: Robotics, 3-D, Close range, Co-ordinate transformation, Six DOF 
ABSTRACT 
Most robots rely upon their rigidity to perform tasks. In the automotive industry the accuracy requirement for operations 
such as spot welding will be of the order of a millimetre. The aerospace industry provides a challenging environment to 
apply robotics in that the accuracy requirements are at least a factor of ten to twenty higher. Conventional robots are not 
capable of achieving this accuracy. However, combining a measurement system and a robot makes use of the ability of the 
robot to move precisely and overcomes the accuracy deficiencies. Photogrammetry provides a suitable method to measure 
the six degrees of freedom of many objects simultaneously. This paper describes work conducted at the Optical Metrology 
Centre, City University and Sowerby Research Centre, BAe Systems. 
1. INTRODUCTION 
Photogrammetry provides the unsual ability to simultaneously measure the six degrees of freedom of multiple objects at 
instant in time. This characteristic can be used in many ways, this paper considers how robotic systems might benefit. There 
are two alternative ways in which photogrammetric systems might be used: with the cameras viewing the robot’s entire 
working volume (Beyer, 1999) or the only viewing the volume close to an end-effector (Clarke, 1999). Both schemes have 
advantages and disadvantages. In the former case a high relative accuracy is required but the robot position can be 
computed in a large area, while in the second case a lower relative accuracy may be acceptable but targets must be placed in 
known positions close to the task. 
This paper describes the development of an end-effector based system that has been integrated with an industrial robot. The 
development of the physical and software components are considered. The physical aspects include: the image processing 
hardware, configuration of the cameras, lighting, coded targets, and location of the targets with respect to the CAD of the 
components. The software aspects concern the methods by which targets are identified, the 3-D estimation of the target 
locations, matching the targets to the CAD information of the components, and estimation of the robot tool centre point. The 
functionality of the system has been demonstrated for drilling and assembly operations showing that the scheme is feasible, 
the current status of the work is designed to assess the capability of such a system to operate within realistic tolerances. 
2. IMAGE COLLECTION 
The image collection process for this project can either take place via the usual frame-grabber approach or using the OMC- 
2D Net system (figure 1). This consists of a real-time hardware processor together with a Digital Signal Processor. This 
system has been described in other papers (e.g. Clarke et. al, 1997, Clarke et. al. 1998, Clarke & Wang, 1999) and is only 
summarised here. 
Extensive testing of the DSP system has taken place and its performance has been shown to be the same as that of a 
conventional frame-grabber in terms of accuracy but with a very low latency and minimal requirements on the PC as 2-D 
target locations are passed to the PC via an Ethernet communication link. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 137