The Robot Pose Error Test Device R&D Project (1983/87)

The goals and objectives were to develop and manufacture a simple-to-use robot position and orientation error testing device, both for off-line as well as for on-line testing and robot calibration and re-calibration purposes.

My analytical / engineering challenge in this project was to develop a mathematical model for static 3D pose testing of robots. I have started this work based on my CNC machine tool testing experiences, as well as based on existing NC standards. (Note, that at that time there were no robot pose error standards; as a matter of fact the result of this R&D project has influenced the tests that were set by the ISO robot pose error standardization committee both in the USA as well as in the UK/ Europe, of which I was a member of for some time).

The method/ model was validated by designing and implementing various test rig prototypes, first using simple contact gauges, with an off-line software evaluation program that I wrote in UCSD Pascal at Trent Nottingham, UK, then later using non-contact sensory devices and the evaluation computer programs that we wrote with my graduate students together at the University of Michigan within the Robotics Research Group in Ann Arbor, USA.

The practical, industrial use and spinoffs included

• several robot tests for industry and academia both in the UK as well as later in the USA;
• several student projects;
• a simplified industrial solution for on-line calibrating the Plessey Beeston/ Nottingham telephone test robot cell's Unimate robot (which won a UK Robotic Prize as well as a UK Teaching Company Prize);
• major influence on the way we were testing robots for the Ford Motor Co. RAACC project in Michigan, and then on
• the ISO robot test standard (already mentioned above).

Major lessons learned include the fact that one has to be able to clearly understand what manufacturers put on their technical fact sheets and what they actually mean by that...

Three photographs below illustrate the first test head design with contact sensors and a FANUC robot being tested. Note, that I have decided to measure 3 locations on the 3 internal faces of the rig to be able to measure position as well as orientation errors, as well as to be able to detect any data reading errors, since the test cube was considered to be perpendicular. Note, that at this stage of the project the data collection was manual, but the evaluation was already software based.

The photo below illustrates the first test rig with a PUMA robot that I have developed at Trent, Nottingham, UK.

The photo below illustrates the close-up image of the first test rig with a PUMA robot holding the test cube that I have developed at Trent, Nottingham, UK.

The photo below illustrates the Plessey Telephone Test Cell, Beeston, Nottingham, UK, that I have designed with my students within a Teaching Company Scheme, with the upside-down mounted Unimate robot and the sensory-based robot calibration device, which was one of the industrial spin-offs of this project. (The cell run for years at 120% speed, won several prizes, as well as created several other R&D and industrial projects both for the company as well as for students and faculty at Trent/ Nottingham, UK).

The photo below illustrates a PUMA robot holding the new, sensory based test rig that with industrial and university sponsorship we have developed at the University of Michigan, Ann Arbor, USA. Note, that I have changed the roles here, in that the robot holds the rig and the test cube is static. Furthermore the new software was evaluating the results on-line, since all data was collected electronically this time.

The photos below illustrate various simple, mechanical robot test devices that my students have designed, built and validated as part of their coursework at Michigan in my advanced robotics course. (The Cartesian test robot we have used was donated by IBM).