DO NOT ENTER! TESTING IN PROGRESS!
I recently penned an article about Thermal Drift in Robotics and it's effect on Repeatability. Unfortunately, I felt compelled to remove the article because it came across as an attack on OEM Robotic Suppliers. This was not the intent. (Reposted article at the end of this post.)
Thermal Drift is an issue that is prevalent in all robotic. Most engineers and technicians are unaware of this issue since it is only relevant in applications requiring extreme precision positioning.
As with all advancing technologies, our defined standards need to improve as well. This was the intended theme of the article. The days of specifications that only show repeatability are near an end. With the merging of AI into all of our exiting technologies, it is only natural robotics will quickly evolve in this area.
There is still much to learn in integration of AI to Robotics but soon, Accuracy will be a specification boldly printing on all OEM robotic advertisements. Temperature, inertia, electronics, and kinematic errors will be overcome by this merging.
Until then, we must continue to identify and understand the cause and effects created by thermal drift and how we, of lesser minds, can meet the challenges of today while contributing to the technology of tomorrow.
To that end, I have continued to work on improving our technology in this area. As with our competition, we continue to work on improving overall accuracy by either adding technology to monitor and track the robot or by compensating thermal position errors through reference target measurements.
Another different approach that I am exploring at this time is through thermal control at the robot itself. I am sure this has been tried and tested by others but I am stubborn in that regard and believe there is a solution to be had.
It's Saturday and as I write this article, I am also performing timed thermal test with our GP12 robot within our lab. (Grandkids are on their way to visit and I only have time to perform a three hour run.)
The hypothesis I am working on today is based on my previous findings over the past several months. We have evaluated data from six robots so far. Two were Motoman GP12s, Two were Motoman MH6s, One Fanuc M10-iD/12, One Fanuc M-10iA/10M, and Kuka provided data for their KR6-R700 sixx.
The following chart represents data provided by Kuka (100% duty cycle/ 100% load)
领英推荐
Although the current robot I am testing today is our demo GP12 unit from Motoman general construction of these robots appear to be the same. All Vendors assisted in our evaluations to some degree, but Kuka provided the most telling data and helped to formulate my current idea. The one issue that stood out the most to me was their ability to predict the drift relative to motor temperature. Of course this was based on a repeated path and without a good predictive model for the entire work envelope and utilization of AI, I don't think we can easily create the kind of results we want at this time.
So my best option for now is to attempt to control the thermal temperature of the motors. This is what I am testing today. The tests actually started yesterday but we are evaluating from various temperature conditions in our facility and the state of the robot.
Yesterday, we tested at similar conditions so this test is just to verify our findings. Our goal in these tests were to see if we raise the idle temperature of the motors and then control them throughout the day, can we eliminate the majority of thermal drift.
To hear more and follow this article, like and link. (Gotta get home to the kids.)
Steven
PREVIOUS ARTICLE
July 28th, 2021 Robot Repeatability
As innovation continues to flourish in robotics, robot specifications are being pushed to their limits. In the field of Metrology, robot repeatability can mean the difference between a $200K or a $400K project. As a company that specializes in these types of systems, I wanted to offer this challenge. If your company makes a 10-12kg industrial robot that can meet their own repeatability specification within 10 minutes from a cold start, please leave a comment below and join our list of verified suppliers. Background… Multiple robot manufacturers claim repeatability in their specifications. My finding is that when you attempt to reproduce the test, (even well below their requirments for load and speed), they fail miserably. The reason for the failure is manufacturers interpret ISO 9283 in a manner that is irrelevant to real world conditions. This is due to (thermal drift) the progressive deviations in position seen during the initial testing for repeatability caused by thermal effects on joint bearings and gearboxes. In most cases this time exceeds more than two hours. From my experience, this is not always a design issue but can be a quality issue. One 12kg robot may be repeatable to 25 microns on a cold start while the next robot may drift 150 microns before stabilization. Robot manufacturers know this and assume you will not attempt to check their claim. If you do and it fails, they simply point to vague statements in the ISO specification to support their repeatability claim and thereby avoid resolving the issue. Regarding ISO 9283… I agree the ISO 9283 document is confusing in the requirements for repeatability since it indicates a variety of tests including Pose Repeatability and Drift of Pose Repeatability. Yes, I understand these are two separate tests. Pose Repeatability is performed in a warm-up state with a five point program cycled 30 times while the Drift of Pose Repeatability test is performed from a cold start and programmed between two points cycled over an eight hour period. Regardless, you can’t ignore drift in your specification since it is an integral part of repeatability and nowhere in the document does it say Repeatability is based on just one test. As a matter of fact, the document does provide a guide for performance based on the application. Machining/deburring/polishing/cutting, Spray-painting, and Adhesive/sealing are the only process that are not included in the Guide for Drift Repeatability testing. If this is a concern of yours, ask your supplier if they can provide the ISO 9283 test results created to support their claim.?
by Steven Baldwin