Industrial Robotics: Robotics and Risk

Home | Articles | Forum | Glossary | Books

AMAZON multi-meters discounts AMAZON oscilloscope discounts

. Risk of the unknown goes together with robotics more often than with any other piece of manufacturing capital equipment purchased by a firm. This section will discuss risk in two different ways. First, the perceived risk associated with change. The change in terms of moving from traditional manufacturing using direct labor and replacing the process with robotics involves risk. The second aspect of risk is related to the unknowns associated with an automation project, either in terms of inconsistent inputs (process raw materials), lack of a sound plan, or utilizing new technology.



The following five categories embody the winning formula based on experience in working with hundreds of robotics users nationwide. The best robotic users in the world are those that:

1) Design the manufacturing process up front for automation. For instance, designing slotted and tabbed parts for a steel assembly.

2) Select programs to pursue where automation offers a competitive advantage. Focusing on product mixes that minimize changeover of process machinery, or parts that require a quality specification.

3) Commit to developing the right team of people on site that accept accountability,

4) Invest in a culture of understanding technology as well as the application of robotics, especially after the system is installed, and having to take shortcuts.

5) Establish a realistic budget for executing the program without First-time potential robotic users play out a familiar scene. First there are the usual suspects of pain that the user has been enduring for a period of time. For instance, the lack of skilled labor, quality problems, and/or the unbearably high cost per piece. Then, during the audit and discovery phase, many of the firms' manufacturing and management team conclude that robotics will provide positive results, but the thought of robotics in place of traditional manufacturing is perceived to be too risky. In the definition of this type of risk, the following short list applies: What will happen if we still need to run the process manually for whatever reason Our firm doesn't have sufficient volume to use robotics because we are a job shop The general fear that robotic technology is too complicated to use The fear on the shop floor where operators see a robot doing a job that was previously done by an operator and begin to wonder if their jobs are in jeopardy All these are valid fears. Section 6 was specifically developed to respond to the second point concerning small-batch runs because that topic is one of the most often perceived obstacles to the use of robotics. So far as the other perceived risks go, there isn't much to argue about because they are subjective and based on emotion.

Education certainly is the right approach to break down perceptions but typically, most firms wait as long as possible to teach employees how robotics can be adapted to their manufacturing environment, or they wait until the pain becomes severe enough. Operators losing jobs because the firm invests in robotics is a myth, because one of the biggest sources of a firm's pain is that they can't find skilled operators.

If a firm is more profitable as a result of its commitment to reducing costs there is all the more reason for it to be able to retain employees. The goal, of course, is to train employees to perform more complex operations, leaving the robot to perform the more redundant work. Additionally, the firm wants to increase productivity with its existing labor. Simply said, firms don't invest in automation so that operators can be laid off. Firms certainly will increase output with less labor, but the skilled labor that was present before the robot was installed still remains the most valuable resource the firm has. Robots often enhance operator responsibilities by forcing the operator to do tasks that are more complex, somewhere else within the firm. Certainly there will be the need to monitor the robot system.

The capital investment for automation is significant, regardless of the amount spent. However there is always an investment, regardless of whether the process to produce something is automated or manual. Many managers forget that there is always a cost to manufacture a product, and they sometimes don't take the time to examine what savings could be achieved with robotics. As noted earlier, the cost of ownership for a manual process is a reality. Along with the perceived risks of robotics, there are certainly risks associated with manual intervention in producing a commodity as well. Some firms feel that the lesser of the two evils is to manufacture the old fashioned way by paying a human operator to pick up a welding torch or a grinding wheel, or stand in front of a machine all day. The challenge for most firms in pulling the trigger to make a capital investment in robotic automation is a function of the fundamental lack of understanding about how their process can be adapted to automation.

For instance, comparing the purchase of other forms of capital manufacturing equipment such as a CNC lathe, a CNC laser and punch, a press brake, an injection molding machine, a stretch wrapper, or a custom drilling machine, the equipment delivers a known technical capability per the manufacturer's specifications. Robotics themselves deliver known values as well, such as reach, number of axes of motion, and weight-carrying capacity. What is interesting is that the user doesn't hedge in the buying decision of a stand-alone piece of capital equipment, other than perhaps the brand and version. There are versions of robotic systems that are approaching the level of product acceptance and maturity as shown by the examples in this guide. The challenge is to continue to educate potential robot users about driving costs out of the manufacturing value stream.

When manual labor is used in an existing or new process there is a whole set of process parameters that can be assumed to exist in order to accomplish the production needs. For instance, why worry about the shopping list of process criteria discussed in Section 2 when the process is manual? The firm need only worry about getting the job done and not necessarily the means. That same thought process is also the Achilles heel of US manufacturing because it’s contrary to lean, and to driving to lower costs. An example is that one of the obstacles commonly cited as to why a firm won’t automate is because they run small production batches. The reality is that we live in a world today where manufacturing must support a batch of one, a production environment that in itself defines Lean manufacturing.

The fact that robotics epitomize the theme of lean manufacturing, due to their inherent traits of programmability, consistency, and adaptability, is something that has to be validated for each user, because it’s not yet common knowledge. The connection with how robotic automation can be utilized to further emphasize small batch production also is not fully understood. The perceptions about automation are often myths and not reality. Unfortunately, the perceptions about what automation can do are also myths when it’s considered that a firm could simply assume that, "robots will solve all my problems". Consider a new program where process equipment must be purchased. The firm quotes the program on a cost per piece basis. The cost per piece is derived from labor inputs, process equipment cost, raw material cost, and overheads. Robotics will level the playing field with low wage competition for the labor component, but will also drive the overall capital investment component, which can be daunting. Firms look at the robotic investment with two sets of glasses. One set provides the view of fear, failure, and the unknown.

The other set provides a sense of comfort because that is the experience of the decision makers. The global market however is not forgiving and has only one set of values; quality products delivered on time at the best cost, and by the way, tomorrow's needs will change in terms of product style and quantity. One of the sayings that is appropriate here, from the CEO of FANUC Robotics' Rick Schneider, is "Innovate or Evaporate". Firms spend considerable funds training their employees in adopting the principles of Six Sigma, ISO, Lean, 5S, and other improvement initiatives. It’s interesting to note that, after engineers attend these various programs the implementation of automation is a natural next step. A firm committed to hiring a Lean engineer or Six Sigma black belt, as an example, should be challenging its manufacturing team about using automation. Engineers and managers will absolutely acknowledge they see the light concerning the value of robotics, and how they can be adapted to complement the manufacturing value stream. From the author's perspective, dealing with an educated user makes all the difference in whether a robotic program will be successful or not. Part of the qualifying process in working with a new potential robotics user is to evaluate the internal capabilities and education, first about their process and second, about their ability to define their goals and objectives surrounding the proposed robotic automation. In effect, a lack of education and lack of knowledge about the initial scope of the work is a major red flag.

Based on many interviews, people engaged in implementing robotic systems would certainly all agree that they have made mistakes that contributed to a robotics system having less than stellar performance, relative to the user's expectations. Not to be too controversial here, but the greatest contribution to poor system performance is usually the fault of the party that intends to use the robotic system. The hot list that follows defines the majority of reasons why robot systems "go wrong":

What were thought to be the product tolerances proved to be inaccurate, so that problems were much worse than expected, and of a random nature;

Specifications were hidden or changed after the scope of work was defined;

The scope of work was not adequately defined at the beginning of the project;

The shop floor personnel simply did not embrace the robotic system and the system was doomed before it arrived on the user's floor process problem The robot was used as a band-aid to solve a bigger-picture The robot system was mis-applied in the first place The system was rushed into production without adequate run. The highest application, with the highest level of risk, was time selected, versus a more straightforward application that had a better chance of success Lack of continuity in key team players from start to finish of the project Lack of a defined budget, causing serious pain later, when the system was delivered, because what the firm wanted and budgeted were not in line Lying to ourselves about how bad a current process or process inputs can be is not a good thing with robotics. Additionally, speculating about the criteria for a new process that is not yet in production will also set up the engineer to fail. Obviously one cannot know everything about a process, nor is there enough time for study.

Whether you're selling or buying robotic automation it’s vital to make sure that the full spectrum of part conditions is put on the table for the team to understand. For example, in a robotic welding application, understanding the extremes of gap condition for a particular weldment will affect how the process is developed to cover the cross- section of process conditions.

Many of the seven risks listed at the beginning of this Section can be avoided by effective communication. There are however some basic principles to adhere to and one of those is to acknowledge the brutal facts and set the expectations accordingly. Given the aforementioned list of the seven deadliest robotic automation sins, there are still countless systems that have been in production for years, despite the technical challenges, because realistic expectations were set. The point of this section is that regardless of a good or bad robotics program, if a sound approach is followed to understand the process, followed by the definition of needs and challenges, then the risks become manageable.

Risk and Cost

Naturally there is a proportional relationship between risk and investment. Most of the first-time robotic users would feel the

"fear" risk of the unknown, in installing a robotic system. That type of risk doesn't correlate to additional cost. The risk that affects cost has to do with presenting an inconsistent input, and converting the input to what would be considered a "good" part. Risk is also a function of taking responsibility for something that you don't control. In other words, a user may need to automate a new program and desire to find suppliers that will take responsibility for making a good product. However, the suppliers don't necessarily control the condition of the raw material, so may not be able to perform as expected, Risk also comes in the form of "what you don't know," and usually appears when the system is already built. The system is being de-bugged for production and there is literally no way to find out about this type of risk until the system is being validated.

Using a conventional dictionary, risk in terms of automation can be defined as many things, but the short list would include the following: The probability that a particular threat will exploit a particular vulnerability of the system There may be external circumstances or events that must not be allowed to happen for the project to be successful. If such an event is likely to happen, then it would be a risk The chance of something happening that will have an impact upon objectives. The risk is measured in terms of consequences and likelihood Risk is everywhere in manufacturing, especially when something new is being developed. One of the greatest strengths of US manufacturing is its ability to develop new products and have a "do what it takes" attitude. This section tackles the "realistic" aspects of risk when considering the implementation of a robotic system. Risk of unknowns is fine, and robotics would not have proliferated in world manufacturing to the extent it has without managing that risk, and having a sound implementation plan.

It’s ironic that the robot itself is the most reliable piece of equipment in any manufacturer's facility anywhere in the world. Of course when a "system" has a hiccup, you always hear, "that robot is a problem". For definition purposes, the risk that is referred to in this section is completely targeted at the robotic system, as including all peripherals that are integrated together to perform a process or task. Robots today exceed 70,000 hours mean time between failures and there are hundreds of robots in the industrial workplace that have exceeded 15+ years without a failure or with minimum maintenance. In summary, risk is associated with the following attributes;

the integrated solution that includes all peripheral

the introduction of raw materials with variable tolerances

the process used in converting inputs to outputs components apart from the robot (inputs)

With the appropriate amount of cost, just about any application can be automated. In the quoting or estimating process of a robotic system, the issue isn't necessarily how the solution can be derived.

The issue is the cost to manage the risk, and who takes responsibility for the cost of that risk. On behalf of system integrators, this point is summarized in two questions, "who is responsible for making a good part, and what is the definition of a good part"? Users often use the term "turnkey" which needs to be defined carefully.

Turnkey and risk can be synonyms because turnkey means process and part responsibility. Turnkey also means suppliers don’t get fully paid until the requirements are met. In the author's experience, most users don’t have the internal capability or resources to take on projects in a turnkey sense. A result of this trend has been the creation of a large network of robotic system integrators who are more than willing to take on turnkey programs. At the same time, risk is then transferred from the user to the supplier, creating a situation that requires strong communication.

 

=========

'Is the end product currently being manufactured successfully with or without robotics’?

'Is this project a repeat project of something already in production?’

The technology to achieve the desired system performance exists or can be demonstrated.

System performance metrics are defined and known.

Raw and finished product exists as well as parts, prints and solid models

Does the system have any special codes [ie UL) or regulations (ie Washdown)

Is cycle time realistic to achieve?

FIG. 1 Risk Gauge for Robotic Automation

==========

There should be a risk gauge. FIG. 1 shows an attempt to describe a measurement tool for risk developed by Tim Lang. The gauge can be completed fairly quickly, and was designed to be used as a barometer for measuring risk versus reward, in terms of the capital investment decision. Just because there is risk does not mean that the project is bad. It’s simply important to identify the risks early on in the project to save time and money later, when typically there are no funds or time to troubleshoot problems. Additionally, the capital investment increases with risk. The gauge describes three primary categories, one of which may or may not be used by the user. The optional category has to do with project-specific items for which the user is taking responsibility. Examples are noted in the chart. If the user is expecting to take no responsibility for the robotic process, then that category would not be used. Another purpose of the gauge is to identify red flags that are not yet defined.

The next Section discusses ways to eliminate or validate these red flags.

Supplemental information for FIG. la is shown below. This list is intended to ask the user some generic questions regarding the project, and determine how the project ranks within a risk/reward map. The map and gauge are certainly not the gospel, but are indicators to challenge the team about whether appropriate costs and risks have been identified.

FIG. 1 a Additional Robot Product Evaluation

The gauge and risk/reward map are applicable to custom-engineered robotic systems. There is also a large market within the robotic industry for standard robotic platforms, regardless of the application type. Standard robot systems are inexpensive, are literally a commodity item, and are good value. At the low-risk end of the gauge there are many standard robotic platforms with many robotic applications that can be immediately drop-shipped, plugged in, and made ready to run.

FIG. 2 illustrates some examples of standard robotics solutions consisting of hardware and software, including the integrated deliverables that typically accompany a common platform. The user needs only to install the utilities and the system is ready to be used.

In its condition as a standard, the system is not applied, meaning it’s not programmed or con figured to do any task.

Robot palletiring ++ packaging platforms ++ Robot polishing platforms ++ Robot welding platforms Robot machine load/unload platforms ++ Robotic Waterjet trimming platforms

FIG. 2 Examples of Standard Robotic Platforms/Systems

===

Robotic Standard Systems Custom robotic systems using existing technologies

Custom robotic systems using new technologies

FIG. 3 Typical Product Life Cycle Curve (PLC) Industrial Robotics

===

For instance, the welding industry has primarily fitted its applications into just a few standard configurations. For any of these configurations, the user or contractor is responsible for integrating a welding fixture, a means of exchanging any fixture for another in a reasonable time frame, programming robot motions and welding schedules, and test/de-bugging the process until the program is ready for production. The capital investment for the standard configuration is low, varying in relation to the complexity of the part and the process. The same goes for the cost of machine tending; the robot gripper fingers, the system to identify raw material location/orientation, and the robot motions, which are all requirements beyond the standard platform. Standard systems, including the hardware side and the value added, generally yield excellent return on investments for users, especially when the robot system can be kept busy for more hours of production.

The examples of standard robot platform applications shown in FIG. 2 are widely accepted and the equipment is quoted and sold like the commodity capital equipment previously described. As in any other industry, the principles of product life cycle apply in the world of robotics, as illustrated by the product life cycle (PLC) curve in FIG. 3. Robotics for many applications have arrived at the maturity point on the curve, although many curves would be required to capture where robotics are, relative to the PLC curve.

For instance the welding market for robotics is more mature than the market for robotic press tending, at least in the US. The interesting dynamic about robots is that they are programmable machines with a very long life in terms of manufacturing machinery. New technology enables robots to be continually upgraded or redeployed, for tasks that are completely different from the original purchased use. Thus, in effect, new technology, especially in robot software, extends the growth period of the PLC curve for conventional robots.

Regardless of a standard low-risk system, the seven sins are still lurking. For instance, FIG. 4 illustrates a fairly simple-looking part. The customer's expectation is that a standard welding system will locate and rotate the cylindrical part in a repeatable and accurate position, to achieve welds that comply with the customer's welding specification. The expectation is that the supplied equipment will make a component from the user's specified raw materials, that will pass inspection, cycle after cycle.

===

Poor weld quality where gap conditions exceeded 2mm.

Weld quality is acceptable with gap less than 2mm 7 1 External laser guided sensor to enable robot to adapt to weld conditions that are outside of specifications

FIG. 4 Visual Differences in Weld Quality and Product Quality when Joint Fit-Up is within Specification

===

At the initial glance the system requirements look simple enough because a standard platform can be applied and robotic welding is mature technology. Now here is the bad news with this turnkey application:

The raw material tolerances exceed the tolerance for successfully welding the weld joint, because the assembly tolerance exceeds half the width of the welding wire diameter, which is the rule for robotic welding.

The assembly is subject to thermal distortion when heat from the welding arc is introduced into the part. As a result, a consistent weld procedure is virtually impossible to achieve because of the dynamic changes in joint fit-up as a function of weld distortion shorter than was initially specified The cycle time requirement for welding the work-piece is The risks associated with this project are fairly severe, relative to producing a reliable, inspection-compliant, robotically-welded work-piece, although there is a solution that will remedy the risk and enable the system to produce an acceptable and reliable product. The fact that the risks and problems occur in everyday implementation of a robotic system is fine. The point that this section is trying to make is that it’s necessary to take the time to identify the risks and unknowns so that contingency plans can be weighed against the cost and project requirements. For the welding example in FIG. 4, the optional methods to overcome the risks and known problems are as follows:

Use a laser-based guidance system to track the weld joint position in real time in order to keep the welding torch properly positioned in the joint Use the guidance system to dynamically adjust welding settings, including welding travel speed, to maintain the appropriate procedure for a given weld joint gap condition weld size and fill-up gap

Possibly weld the joint in two passes to ensure appropriate Program the robot to weave the weld deposit to bridge the fit

Have the operator fill in the gaps manually, where the gap condition exceeds the wire diameter prior to robotic welding

Depending on which option(s) work, the engineer needs to weigh the value of utilizing them to ensure that a reliable work-piece is produced at every cycle. The consequences are as follows: Additional cost overall in programming and de-bugging the welding, sensor, and robot programs to accommodate all the possible weld conditions caused by random part fit-up Additional cost in welding wire (filler metal), if the weldment must be accomplished in two passes Additional weld time (reduced productivity), if the travel speed has to be reduced, or the weld deposit has to be weaved Additional operator intervention within the process if operator preparation of the work-piece is required, prior to robotic welding Additional complexity, training, and maintenance, due to the use of a laser guidance system

 

====

The cost of automation The benefits of automation Criteria in Balancing Investment vs. Capability Operator intervention within the process (some or none)

- Changeover from one product Shop floor resources to support system (level of training, and personnel capability)

inspection or process validation/feedback to another ( automatic or manual ) (robotic or manual management)

- Compliance tools for converting * What does the firm take responsibility for varying inputs to consistent quality outputs ( adaptive tools ) and pass onto others

-Adapting to potential future projects ( expandability, mobility ) Level of flexibility to handle product mix (more or less flexibility)

FIG. 5 Judging the Robotic Investment Versus Robotic Benefits

 

====

The saying, "it is what it is" seems appropriate here. The condition of the raw material (input) coming into the process does not allow the process to be successful, so adjustment and additional decision making are required to ensure a consistent output. Because of these requirements, the project cost will increase, but at least everyone will understand this, and a decision can be made to move forward, knowing the risks.

The risks and unknowns are determined by going through the shopping list outlined in Section 2. The purpose of the shopping list was two-fold. The first was to develop the project design criteria.

The second purpose was to identify risks and unknowns that affect costs. Most often there is a solution to convert poor inputs to acceptable outputs. The heart of the matter is at what cost? One can do anything with automation, it's only necessary to spend money. Of course, if the return on investment, or the benefits of the investment, are not present, why do it. This is why every project is not a good candidate for robotic automation.

FIG. 5 is a simple illustration that addresses the core of the decision-making for whether a project should move forward or not.

Unfortunately there is no universal formula to weigh a decision on a good or bad project. Cost, risk, and capability of the robot system, need to be compared on a case-by-case basis. As mentioned several times in this guide, cost and risk are directly proportional to the agility of the robot system. The definition of agility is that it consists of a combination of flexibility, adaptability, and the level of decision making required of the system. There is a balance between the level of agility and the investment. It's easy to see diminishing returns on the automation investment when the equipment is required to take on more process responsibility with little incremental economic gain. Vice versa, the robot system has to be able to deliver a wide set of capabilities to make the investment worthwhile.

===

Case height can 7" from one case next to another

Cases are bowed and not uniform L Leakage from the product deteriorates condition of cases / staggered layers of cases

FIG. 6 Example of Robotic Compliance Requirements in De-Palletizing Non Uniform Unit Loads

===

FIG. 6 illustrates another example of a situation where additional cost is required to convert poor input to acceptable output.

Examination of the stacking arrangement of the cases on the unit load shows that each case varies in position in the X, Y, and Z planes, and there is no relationship between one case and another.

The project expectation is that cases will be picked up by the robot from the incoming unit load and restacked on a new pallet within a certain time frame per unit load. The risks are as follows: The cycle time requires that multiple cases be picked up in each cycle

There is no rhyme or reason relative to the positions of each individual case

Some cases are warped and some are not warped

FIG. 6 shows that each layer of cases is staggered

The bottom and top halves of each case are held together

with two plastics straps. Sometimes the straps are missing, or there is leakage from the items in the case, causing deterioration of the case condition when picked up.

This application requires the robotic system to have the ability to identify the position of each layer of product, to establish a frame of reference for that layer. Additionally, the robot gripper must be designed to allow movement in the Z direction to suit the varying topography of the cases in a layer. To make the specified cycle time, the entire layer is required to be picked up at each cycle causing further risk and challenge because the gripper itself has to comply with a layer of cases that are all at varying elevations.

At the left in FIG. 7 is shown some iron castings, each of which has a riser that is located in a common position, left over from the casting process. The user's supplier of the castings tries to grind the risers down to a nominal offset from the outer diameter of the base material, but with a manual process the riser height still varies within a tolerance off 0.188 inch. The casting is to be loaded into a vertical lathe for machining. As seen at the right, the machine has a three jaw chuck to grip the casting OD. The risk that results from the poor input is as follows: If the robot locates the casting where the three jaws can grip on the outside surface where the riser is located, a mis-load could result if the riser is too large. A misload could damage both the part and the machine. Certainly there would be downtime

===

Work-holding in the CNC vertical lathe (three jaw chuck)

Riser on the OD of the casting (variable riser height every casting) Cannot chuck on riser or machine is down FIG. 7 Machine Load/Unload Example

===

This application requires the ability to detect where the riser is located on the casting at every cycle to enable loading to be done safely. The consequence is additional cost in terms of the technology to enable the riser to be detected. The user can't rely on an operator to place the castings with the riser orientated in an exact position when the operator is loading the system with raw castings.

FIG. 8 illustrates the need for many variations of material removal applications. The technology of adaptive control through the robot or a peripheral device is widely accepted for material removal applications because of the frequent part-to-part dimensional variations, as well as variations in quantity size of material to be de-flashed or trimmed. For instance, in the plastics blow-molding process, the parts shrink while cooling, which adds a whole other twist to successful de-flashing / trimming of the work-piece. Of all the robotic applications to pursue, material removal should be a priority from a standpoint of achieving consistent quality, reducing work-related injuries, increasing throughput, and labor savings. These jobs are among the most manually intensive, and, unfortunately for the firms concerned also have the highest quality requirements.

Material removal (finishing operations) tends to be one of the last processes in the value stream, prior to the customer receiving the product. There is a great deal of standardization in robotic material removal, from polishing and buffing, to de-flashing. The standardization in flexible devices and robot software technology that enables systems to remove material and to achieve uniform surface edges has resulted in lower investment costs for the user. That is the good news. The risk in these projects is to do the development work in choosing the right auxiliary device, the best media for removing material, and the trial-and-error process programming development to achieve the consistent surface finish that meets specifications.

Additionally, the firm will require a talented technician to manage the system on the shop floor on a daily basis.

The balance is in programming all these components to make a good part, otherwise a scrap part will result. Imagine the challenge of deburring transmission components such as high-alloy gears that cost thousands of dollars to make. On the one hand the operator introduces a high probability of making a mistake and scrapping a gear, which further justifies the need for a robotic gear de-burring system. On the other hand, the cost to develop a robotic system may involve a return on investment longer than the ideal. The bottom line is, what does it cost the firm to scrap a part? The robot system serves as an insurance policy for preventing a potential future problem.

For material removal applications, unless a robotic process is developed and tested around a specific part type and process, there is no way to know whether the process will work. Thus, for material removal applications where, if the identical process is not presently being done elsewhere, an engineering study should be performed to bring up red flags and then plan around the cost of risk. It’s one thing to see something similar in a material-handling application, but an entirely different thing to see a similar material removal application. Similar still means not identical. Section 4 will cover material removal applications as the subject matter for developing an engineering proof of concept, prior to making capital investments.

===

 

FIG. 8 Examples of Material Removal Applications Requiring Adaptive Control through Robot or Compliance Device

===

The several examples mentioned previously describe the classic struggle of using a robot that by its very definition, will do the same task the same way, over and over, except that this time we want the robot to adapt its cycle in terms of converting variable-input conditions into consistent, workpiece-compliant outputs. Robot technologies have now enabled the robot system to do exactly that task, which is a wonderful thing. The shopping list can be used to expose risks and unknowns. Unfortunately, process risks don't become visible until the system is built and in the de-bug mode. At this point, additional cost to manage a hidden risk is not a good thing. At a minimum, if the engineer knew the risk early in the audit phase, the budget could be adjusted for the additional cost. Hidden risks equate to hidden costs, after the fact. This situation is where the party providing the turnkey solution, and the customer for the turnkey solution, battle over who is responsible for the costs to overcome the challenge. Integrating and building a robotic system is fairly simple. Making a good part over and over, within the overall throughput requirements, is where you earn your stripes.

Hidden Risk No one likes surprises that cause additional work or costs in any endeavor, and especially not in robotic system implementation.

Even though there are surprises, in a 15-year career the author has seen only three (3) systems that were built and never shipped, and that is over hundreds of robot systems. All three of the projects worked, and what the robots were doing was impressive. The problem was with specification criteria that the robot system could not meet without significant additional funding, which was not available, so the projects were cancelled. There have been problems with other systems, but with the exception of these three, everyone worked through the issues and the systems continue to run today.

Any organization that is focused on implementing custom-engineered systems will have problems. There are simply too many details, and until the system is built, the process cannot be debugged completely.

An interesting story about hidden risk has to do with a turnkey project for welding the handle of an automatic hand clamp, used by construction workers. The hand clamp was a very popular tool and was advertised on television because of its ease of use. The robot system was purchased to weld two sections of the clamp together along a butt seam. There was only one weld per clamp side, and the weld bead was about 3/16 inch wide, and 4 inches long. A standard system was to be used, and welding tests had been accomplished prior to the system being approved. The expectation of the user was for the system to produce a reliable and acceptable part.

The interesting thing is that, as the system was being installed the user learned that the parts had to pass destructive testing, where the base material had to fail before the weld, and the base material was known to be made of high-carbon steel. The logic for selecting the welding filler metal during the project implementation phase was to match the filler metal to the base material. This decision to match the chemistry of the filler metal to that of the base material was not a good idea. Sample after sample failed the destructive test.

The user was in a position where, if the system could not make a testing-compliant part, then costs would escalate and the user could lose the contract with the customer. The battle lines were drawn.

There were two opposing arguments. The provider of the turnkey solution was required to produce a "good" part, and should have known that a different filler metal would have to be selected to achieve the mechanical properties in the weld that would withstand the destructive testing. The other side of the argument is that, had the user defined the parameters adequately, or even had known that destructive testing was a requirement, the situation could have been avoided. The arguments are to an extent irrelevant because in the end, the process has to work, and without a solution, both sides lose.

The conclusion was to select a new filler metal that would allow the weldment to achieve a much higher level of toughness, and as a result the base material would fail in a brittle condition before the weld failed.

Other examples of problems are listed below:

- Adding a chiller unit to a washing system after a machining process, because, during the washing cycle, the die cast aluminum part was heated excessively, causing thermal expansion problems for the next machining process,

- Finding that an attribute of a product (input) changes after the fact. An example is that, during the validation phase of a project it was confirmed that the loading and unloading of an iron casting into and out of a special machine fixture was consistent. The engineer then specified those surfaces as suitable for the robot to grip the part so that it could load the machine. However, the engineer was lucky on the day he was testing the process manually, because the castings were consistent on that day. The selected gripping surface was later found to have a very broad tolerance, and as a result an accurate loading process was no longer valid.

- Palletizing bags of seed, where the bags would be presented to the system with a broad distribution in terms of where seed settled in the bag. A bag flattener and weighing system were added to ensure that uniform bag content and shape were presented to the robot, prior to the palletizing sequence.

- Finding an interference or access problem between the robot gripper and a location that the robot needs to reach, after the system is built. For instance a welding fixture prevents the robot from reaching the position on the part for welding, or there is not enough room for a robot arm to reach inside a machine tool door to unload and load a part to be machined.

- Exceeding the time you thought it would take to program, especially with process-intensive applications such as welding or material removal. Teaching the robot where it needs to move to in space is straightforward. Making a good part is different. For instance, achieving the correct weld attributes, or the specified surface finish of a chamfered, machined, edge after deburring, are examples of trial and error tasks.

- Obstruction of a machine tool chuck by metal chips or debris, causing problems with the robotic loading process.

- Finding out that humidity affects the properties of raw blocks of plastics polymer that a robot system is de-palletizing. The blocks stick together in conditions of high humidity as the robot moves blocks from a pallet to a conveyor that transfers the plastics to a mechanical shredder.

Risks of Your Own Managers and Personnel

A company that produces equipment for the agricultural industry finds that its biggest obstacle in achieving good productivity is proportional to how the shop floor operators decide they're going to work on a given day. This company uses welding robots, and the systems rely on the operator feeding the system. The operators in this situation affect productivity, with or without automation. Without automation, productivity would certainly be significantly lower, but the firm struggles to achieve the uptime it knows the robot systems are capable of producing.

Another company sells machining time and fits a classic job shop model of producing something different every day. This company struggles with the risk of their customers keeping their commitments to product volumes. In other words, their customers could pull contracts or adjust contracts for part volumes in an instant. As a result, it’s difficult to manage people resources and capacity planning. Hiring and then laying off employees doesn't seem to work well for any long-term continuity. Additionally, the firm has to be very careful about long-term investments that limit its ability to react to customer needs. Each investment has to have inherent flexibility as part of its core requirements. The definition of automation includes words like re-deployable, consistent, flexible, and adaptable. Automation allows the firm to stay agile, which is the key for manufacturers regardless of fitting the job shop or OEM model.

Product lines continually change, as do customer demands.

Whether automation is the solution to this long-term problem is hard to answer, but robotics will continue to become more flexible and easier to use, allowing users to adapt to the ever-changing business climate.

Ryan Burg of The Department of Legal Studies and Business Ethics at the Wharton School of Business brings up an interesting point of view about manufacturing: "There is considerable debate in academic management over the question of why American manufacturers have done such a pitiful job at lean manufacturing implementation. My view, and I think I am not alone, is that lean manufacturing and the Toyota model are better grown than built. Lots of complex systems are better grown than built: operating systems are a good example. Windows XP is a pretty good operating system now, a mere ten years after it was designed to be released. Mac OSX is a good operating system, built with a UNIX backbone, parts of which are decades old. When you need complex order, it has to evolve. Lean systems (kan-ban, just-in-time, inventory-free) are not in themselves productivity inducing. What they actually do is make it possible for processes to be adjusted more actively ("constant improvement") so that the system will be able to evolve. There are moments when clean slates are useful, but they can destroy as much value as they bring. The best laid plans can fail because missed details accumulate. The trick is, therefore, to design in a way that can be adjusted, or at least it seems at first glance. However, when you look more closely, I think you find that the REAL trick is to have people who actually care to make the adjustments and are empowered to do so, people who work in collaboration to achieve the constant improvement. This is the part that America, and the contentious politics of our labor have really struggled to work through because the best manufacturing practices in themselves don’t make anything." The best automated system designed to be completely fool-proof still requires support. The point is that people will always make the difference in how effective manufacturing processes will become, with or without automation. The ability of talented caring employees, who have the attitude to make positive changes must not be underestimated. These people must be empowered to make decisions.

Managing What You Control

New users and even existing users to some extent, will want to know if their project is being automated somewhere else. Any first-time user should take a road trip to see automation in a production environment, and chat with all the various members of the team from maintenance to the owner. Education is the best practice when pursuing a first-time project.

What does a firm control when it decides to move forward in evaluating automation: The user understands the customer requirements in terms of the product specifications that need to be delivered Product volume and delivery requirements to the customer.

Albeit this category could be completely undefined, or demand is on a daily basis and the only rule is that change will happen Raw material cost to produce the product in the specified volumes and time frame The process or sequence of operations required to make the product How much capacity in process machinery and labor are needed to produce the product Amount of profit per part Process requirements to convert (raw material) inputs to (finished product) outputs An educated user will understand these categories and by identifying risks within them should be able to communicate them throughout the team. In terms of automation the user controls: The decision to build the robot system internally or externally Specifications for the automation requirements, defined in a document labeled scope of work How the system is designed isn't really the user's concern, other than that the system has to perform to the requirements in the scope of work. Imagine then, that there is no scope of work, or the scope of work is poorly defined to meet the project expectations. The scope of work is discussed in Section 5 and it’s truly a make or break process in terms of controlling costs and risk. Requirements and personnel always change, but those changes don’t mean that the scope of work is disregarded. Ultimately users control the level of definition for someone else internal or external to the firm, to deliver something. Getting back to the standards that users buy, in terms of the commodity capital equipment, including the robot itself, there is no issue here. The risk is all with the level of definition laid down for a custom system, between the user and the supplier.

Risk of Technology

Despite the maturity of the robotic market, the inherent flexibility of robots enables new technology to be continually developed, as well as new applications being created with existing technology.

There are two concepts regarding technology and risk. First is the risk associated with a new robotic application that is unique and can't be compared to any other application, anywhere else. Second is that new technology may be validated in the development lab but not necessarily in the production environment. Either form of risk will yield a high score on the risk gauge, but that is not necessarily a bad thing. There would be a higher cost associated with the implementation due to lack of experience for either event. Cost is relative though, to the specific benefits that can be gained.

Any investment in a robotic system today will yield significantly greater value in terms of cost to functionality ratio, and reliability than ever before. Robotics themselves have been inherently improved in terms of reliability and ease of use. The reliability and ease of use have certainly decreased the risk of a robot investment.

Earlier text mentioned proof of concept, which will be covered in Section 4. Some applications that should be further qualified with proof of concept to avoid hidden risks are as follows: Dynamic robotic sorting of laser-cut blanks. This technology is applicable to any CNC-based cutting table (waterjet, plasma, laser, or router). See FIG. 9. This sorting process can be a huge manual bottleneck in the value stream.

Robotic random bin picking, eliminating the need for an operator to sort raw material for the robotic system. See FIG. 9.

Robotic force sensing for assembly. See FIG. 10.

Adaptive robotic welding to accommodate varying weld joint volume. See FIG. 11.

Adaptively "flexible" robots coupled with trimming, routing, de-burring, grinding, and polishing tools, designed to remove excess material such as parting lines and flashing on steel, composites, and plastics components Palletizing mixed loads of products on a single pallet. See FIG. 12

===

1. Dynamic Robot Part Sorting Robotic de-nesting of cut blanks from a CNC cutting table (i.e. plasma, waterjet, laser. router). The robotic system is integrated to the CNC nesting system to receive positional data to pick cut blanks from a master sheet including the remnants. Each cycle, the master sheet could have a completely new nest in terms of the recipe of shape styles, location, and orientation. There would be too many possible combinations of recipes so every cycle requires dynamic robot paths created from nesting information about each shape attribute. Other challenges are handling parts that are tabbed to the master sheet.

2. Robotic Bin Picking Robotic bin picking of randomly located work-pieces in a basket, bin, container, gaylord enables the user to eliminate the manual staging of raw material to the robot system. Unmanned production is the direct benefit and for material handling systems the benefit is significant savings.

Challenges include picking the work-piece with literally any orientation, parts that are interlocked with each other, avoiding the container sides, robot path planning into and out the bin to pick randomly located product. Another challenge is planning for safe retreat out of the bin in the event of a fault condition FIG. 9 Examples of Applications where Proof of Concept is Useful to Avoid Hidden Risks Force Control Applications FIG. 10 Examples of Applications Where Proof of Concept is Useful to Avoid Hidden Risks

===

Adaptive Welding (through the robot and peripheral sensors). Locating the weld joint. Real time robot path correction. Adaptive fill for varying weld cross section and gap condition Adaptive process parameters (weld travel speed, weave parameters, and weld procedures).

Decision making about number of weld passes to adequately fill a weld joint. Part verification prior to welding Ideal where work piece tolerances exceed the ability to use nominal weld procedures in order to produce a quality product; Casting to casting weldment Variation in forming, cutting process Stack up of tolerances from the pre weld Effects on the weld seam tolerance as a assembly process function of thermal distortion

FIG. 11 Examples of Applications Where Proof of Concept is Useful to Avoid Hidden Risks

===

Mixed Load Palletizing

- Multiple product types (skus) on a single unit load Handle multiple case sizes on a single unit load

- Build a unit load that is stable without control of when specific cases are presented to the robot Multiple customer orders on a single unit load

Insuring the robot palletizing is not a bottle neck within the overall product flow FIG. 12 Examples of Applications Where Proof of Concept is Useful to Avoid Hidden Risks

===

The trend here is that robotic control technology is continually being developed to enable the robot to make dynamic process decisions based on a set of varying inputs, while converting them into an acceptable output. The robot, in a sense, is an intelligent and agile manufacturing tool that can adapt to changing conditions. The risks are the complexity, setup time, and maintenance of these tools.

Robot technology will continue to find ways to mimic human dexterity and decision making. The reward is cost savings to the user because the robot is essentially adding more and more value to the process. Finding that balance between capital investment and robotic agility will still be the right thing to do for the user.

Without the important adaptive tools and software technology behind these tools, the market for robotics would be severely limited. The standard platforms are great but their growth is limited as well as their applications. Continually-changing technology is a good thing if you know the risks, and the proof of concept discussed in the next section plays a role in that technology- validation phase.

 

Top of Page

PREV   Next | Article Index | HOME