Robotics Online - RIA - Robotic Industries Association
Featured Articles

Reflecting on Robot Safety

by Jeff Fryman , Director, Standards Development
Robotic Industries Association

As the Robotic Industries Association (RIA) 'point man' for robot safety, I have had the pleasure of meeting and working with a number of robot users in plants throughout North America. In many cases, this has been as the instructor conducting the Robot Safety Standard (ANSI/RIA R15.06-1999) course at an in-house training seminar. This is just one of several courses the RIA offers as part of its industrial robot safety training program. In-house training seminars are one of a variety of robot safety resources the RIA offers to the industry. 

Starting in 1999, these courses were intended to introduce the 'new' robot safety standard to the people who needed to know about it and comply with its provisions. New because it replaced the 1992 edition of the standard and contained new and important safeguarding strategies and guidelines. Since the last of the phase-in deadlines passed in June of this year (2002), it has been a little hard adjusting to describing the R15.06 document as 'the' robot safety standard, no longer the 'new' standard. This transition in thinking has allowed me to reflect on what some of the more important issues may be going forward.

One question that I am repeatedly asked is, 'Can I use my old robot in a new work cell?' The answer is a definite yes, with a number of caveats. Sorry, but it is not straight black and white - it has lots of color! So… let's do some color commentary (it must be football season).

First, as background I would like to comment on the thoughts of the committee as they drafted the current standard. I believe the committee was very sensitive to the needs of industry, including consideration of cost issues. They acknowledged these cost impacts by ensuring that there was a valid return on improved safety for the investment cost in implementing new requirements. Safety has a cost, but a sound safety program would look like a good investment compared to the costs of litigating a serious accident.

'Grandfathering,' or the requirement for retrofit, was the single most contentious issue in reaching consensus on the standard. One camp did not want to look back, the other camp felt non-compliant installations were not entitled to another 'free ride' for having done something wrong or inconsistent with the previous standard. Both camps agreed that the provisions of the '99 standard provided improved safety and should be required for all future installations.

How did this play out? Basically the committee looked at two aspects of a robot system, the robot itself, and the work cell.

The robot hardware itself is 'grandfathered' from retrofit (though voluntary upgrades are allowed). This was a concession that hardware is difficult to change, and may not be cost effective. An example of this would be the provisions for limiting devices, particularly hard-stops, on some of the axis. Though not difficult to include when designing the robot, installing hard-stops could be difficult to impossible on existing hardware.

On the other hand the system or work cell was easier to change, and over time typically is changed and upgraded in the normal course of business. Thus, safety enhancements could be made incrementally and their costs amortized over the useful life of the robot application. Typically a robot work cell would have a useful life on the order of 5-7 years before its technology (albeit not the hardware) was obsolete. Cost depreciation and amortization would allow for an orderly and economical replacement for older systems. Cost tradeoffs for new, remanufactured, or rebuilt robots would be part of the business case made when new systems are designed.

How is this implemented in the standard?

Clause 1.3.1 New or remanufactured robots, provided for a grace period of 24 months intended to allow the manufacturers time to re-engineer their products and clean out existing inventory of the older non-compliant models. Re-manufacture-the changing of the capabilities or functionality of the robot with new and upgraded capabilities; i.e. new controllers, stronger servo actuators and drive train to increase payload, etc., including re-engineering by a third party-is differentiated from rebuilding-the complete overhaul and refurbishment to original specifications, albeit with newer components and latest software releases. The business decision to invest in remanufacturing a robot versus rebuilding a robot includes the decision to accommodate updating the robot to the new safety requirements in the standard.

Clause 1.3.2 Rebuilt or re-deployed robots, is the final part of the robot hardware equation; the robot hardware is 'grandfathered' to the requirements of the standard effective on the date of manufacture. Safety enhancements are allowed, and encouraged, but it can be an 'ala carte' selection-those features that are easy or economical to do rather than the whole list.

Re-deployed robots can be re-deployed from one cell to another cell in a similar application or can be re-deployed to whole new applications where only the end-effector (complete with supporting hardware) is changed, and the task program is changed. Now this is what is allowable, not necessarily feasible given the specialization of some robots. For example, a robot can go from one MIG welding application to another MIG welding application but on a different part. A robot can also go from a welding assignment to a material handling application by changing the end-effector and the task program. This may be useful in utilizing older robots which no longer have the repeatability tolerances necessary for precision work. An old robot with a repeatability error of 2-3mm may work just fine in a palletizing application where the carton stack would not notice a 3mm error compared to a welding application that could result in a total bead void in the same circumstances.

So now the conundrum. Since new work cells must be in compliance with the new standard, using older 'non-compliant' robots in new cells may require some compensation in the system installation to provide the appropriate level of safety. Again, depending on the age/design of the robot, some of the additional safeguarding requirements may not be practical, even though it is allowed for in the concept. These are all considerations that must be looked at in the business plan to decide on the design of new work cells and the components that go into them.

The biggest, but not only, issue may be the necessity for a dual run chain to meet the requirement for 'control reliable' circuitry. Most older robots only had a single run chain. The good news is an external run chain with monitoring and an additional set of motor contactors can be added to satisfy this requirement. Of course, this may be much easier said than done. Control reliability will be needed if you follow the prescribed methodology in Clause 8, and may be required based on your determination of risks from a risk assessment following the procedures in Clause 9.

Another issue could be the lack of limiting devices to establish the required restricted space. This may necessitate a larger footprint for the work cell to achieve the necessary clearance, or applying additional safeguarding is required.

Selection of certain optional operational features may also require accommodation. Using a teach pendant inside the safeguarded space requires an enabling device, but not necessarily the three-position enhanced enabling device. The use of high-speed attended program verification will require an entire list of specific compliance items including the enabling device, control reliable circuitry, and adequate clearance.

I try to answer these and many other questions in our in-house seminar offerings, and in our annual robot safety programs, the spring regional and the fall National Robot Safety Conference, this year in Ypsilanti, Michigan, October 21-24. See our web site for full details on the safety conference and the other robot safety resources available to you through the RIA.

Jeff Fryman may be contacted by e-mail at; or by phone at (734) 994-6088.

Automated Imaging Association Motion Control Association