Metrology Solutions for Adaptive Robot Control

A robot guidance application from ATT Metrology integrated activities within a custom process workflow

Published: Monday, July 10, 2017 - 11:01

Sponsored Content

Opportunities are becoming more frequent to apply metrology to adaptive robot control for many applications. There are many different techniques to accomplish this, from regular calibration to real-time feedback, using anything from laser trackers to optical devices in order to capture high-accuracy position data for these processes. ATT Metrology recently provided automation to complete an adaptive robot control (ARC) system designed to install large glass panes into custom window frames.

The system integrated a laser tracker into a robotic build process that utilized a large industrial robotic arm. Robotic arms are known for their repeatability, but accuracy in high precision is difficult to achieve without an outside system to guide the robot’s end effector to known positions. To achieve better accuracy, we created a system that measured corner cube targets on the end effector to obtain a location frame at key positions in the process, then provided a frame-to-frame transformation from the robot’s reported position to the measured one. This allowed the robot to make fine adjustments in the end effector position to coincide with the target location.

Figure 1: ATT Metrology’s robot guidance project designed an adaptive robot control (ARC) system to install large glass panes.

Project overview

In order to create the ARC system, ATT Metrology needed to provide automation programming to communicate with the robot controller program for reading and transmitting data to and from the robot. We developed a communication protocol for the process that ran via a PLC controller database, which provided end-effector position data as well as distance readings from laser range-finding devices embedded in the robot cell. We also created software to drive the laser tracker to the end-effector corner cube targets and record measurements. Finally, we solved for the measured frame location and calculated the transformation to the measured frame from the robot’s received position data, and transmitted this back to the robot controller program over the PLC communication channel.

The high-level ARC process can be broken down into four major steps as depicted below.

Figure 2: High-level process for adaptive robot control (ARC)

The automation of laser tracker operation and kinematic calculations was developed using ATT Metrology’s Automation Framework, which allows for integrating activities within a custom process workflow. This modular approach, combined with a workflow editor and previously developed activity modules, allows for assembly of automation procedures, often without the need for additional programming, only a customized configuration. However, for this particular project, we did need to develop a custom activity for the ARC process as a “plug-in” activity, which we were then able to configure within the automation framework workflow.

The overall system schematic showing how the various components were related and connected is shown below.

Figure 3: System architecture for the ARC solution

The robot is a Fanuc M-2000iA six-axis arm capable of lifting 1,200 kg, with a vertical range of 6.2 m and a horizontal range of 8.3 m. The kinematic package we used was New River Technologies’ Spatial Analyzer (SA). This package also provides connectivity to the Leica AT401 laser tracker and has a software development kit (SDK) that allows third-party software to communicate with SA and the attached tracker. For PLC communications, we used a PC-based communications package called NetLogix. The modules for communicating and integrating these components were developed using ATT Metrology Advanced Automation software.

Robotic arm accuracy vs. repeatability is a frequent topic of conversation when evaluating the feasibility of applying arms to certain applications.  The absolute position accuracy is the ability of the robot to reach a specific programmed position with a minimum of error. This can be thought of as the closeness that a centroid of certain attained positions can attain in relation to a desired target position. Repeatability can be defined as the closeness of agreement between several positions reached for the same directed position.

So, accuracy is how well we group around a target position, while repeatability represents how closely our “shots” are grouped. The diagram below shows how these two qualities of positioning related to each other:

Figure 4: Positional accuracy vs. repeatability

According to Fanuc engineering specifications, the particular arm model used in this solution is capable of accuracy somewhere in the 1–2 mm range, while its repeatability is somewhere in the 0.5 mm range. We were seeking accuracies in the 0.1–0.3 mm range to meet the accuracy requirements for this project. The arm is capable of making fine motions in this range from a particular starting point, but it needs a direction (i.e., vectors) to move along toward a given target.

Another issue in getting an arm to a guided location concerns mechanical issues, such as “backlash”—i.e., the lost motion taken up by small clearances between elements in the arm’s drive mechanism, such as gears. The robot control program was responsible for setting up a motion to an initial location such that backlash was minimized; this allowed for fine-grained motion adjustments along the same general direction, which minimized the effect of backlash.

The combined effect of the laser tracker measurements and positional guidance combined with the robot arm’s natural repeatability gave a resulting positional resolution of both high accuracy and high repeatability.

We decided on a system architecture for the communications built around the IP protocol. This allowed us to break the system into parts that were each separately testable. The laser tracker and kinematic automation simply received data packets over an IP connection. This allowed us to send test data to the program to test calculations and to log communications for fault tracing and operator feedback on an MMI installed on the industrial PC that was running all the automation control software.

Figure 5: Laser tracker and kinematics MMI

The communications software also provided operator feedback and could be used to evaluate data coming from the robot control program over the PLC controller “tag” database. The values are grouped according to data that come from the robot controller program, and those values that are returned by the laser tracker and guidance application. As part of the protocol, the robot controller program sends a position and step or sequence number, along with optional distance data from the laser range finders in the frame assembly rig. These data are then forwarded by the PLC host program to the laser tracker control software over the IP network that connects the various components.

Figure 6: ATT PLC communications host/MMI

Conclusion

The system proved capable of meeting the accuracy requirements given at the start of the project and showed that the combination of metrology techniques, modular automation tools, and robot control made a powerful combination when used in an application that required both high accuracy and high repeatability.

If you are interested in projects of this nature and want to learn more, come by and visit me and the ATT Metrology team at next week’s Coordinate Metrology Society Conference in Snowbird, Utah. You can find us in booths 112, 114, and 116.

About The Author

Michael O’Shea

Michael O’Shea joined ATT Metrology as chief technology officer in April 2015. He brings more than 25 years of experience building and leading the development of software solutions with applications in a wide variety of fields such as HR, aerospace, and financial services. Most recently he was manager of software development for Russell Investments. Prior to joining Russell in 2007, he was western regional director of software development at Logicalis serving a long list of clients in telecommunications, biotech, and a number of Seattle internet startups. He was a software engineer at Boeing for 10 years from 1989 to 1999.