PickFlex: Vision Guided High Speed Pick & Place
by JMP Engineering
JMP Engineering Posted 09/21/2012
JMP’s customer came to them with an ergonomic issue in their plant. The process involved workers manually grouping the ice cream bars on a high-speed flat belt conveyor and placing them in the trays on a tray conveyor. From this tray conveyor, the bars were pushed into boxes and sent to downstream automation for packaging, labeling, and finally, shipping. The trays on the tray conveyor had five different pack patterns: 1-pack, 3-packs, 4-packs, 6-packs, and 9-packs; the line ran at a rate of approximately 150bpm (bars per minute).
The main issue with the line was the high speed, repetitive task of packing the bars in the 1-pack trays. Workers stood in one position, with their heads bent over the conveyor, reaching in front of them to orient the ice cream bars on the flat belt conveyor and then pull them into the pockets of the tray conveyor. Due to the geometry of the tray conveyor, this meant the workers were packing bars onto a conveyor moving at a blistering 30in/sec! Clearly this was a very taxing job for the workers, who could only stay at that position for 15 minutes before rotating to a different job. Even after only 15 minutes, the workers would walk away stretching shoulder and back muscles to try to alleviate the strain.
Drawing on their expertise with robotics, a solution was designed that would use three food grade robots rated for high speed continuous duty picking and placing. These robots came with a proprietary software package from the vendor that provided the high speed picking and placing algorithm as well as line tracking capabilities. Simulations of the proposed cell were developed, and based on the results from these simulations, the concept was deemed feasible and was planned to operate at 99% OEE. This target meant that the cell could only miss one bar per 100 coming down the line.
The system was built and commissioned on JMP’s floor. Sample product was brought in from the customer to ensure the robot tooling was capable of handling the parts, and extensive troubleshooting was done to ensure the tooling was as reliable as the continuous duty robots they were mounted to. At this point the system was deemed to be capable and ready for install.
Once the cell was shipped, the project team came face-to-face with many issues that had been either unforeseen or assumed to be manageable on-site.
The fundamental challenge in this project was the speed of the product coupled with the variability in product flow. This affected the robots the most on two-pack patterns in particular – the 1-pack pattern, where the tray conveyor was moving at 30in/sec, and the 4-pack, where the robots were restricted to picking only two bars at a time instead of the usual three. Because the robots were picking only two bars at a time, a larger percentage of their time was being spent moving back and forth between the in-feed and out-feed conveyors, effectively forcing a reduction in the robots’ efficiency.
The variability in the flow caused by mis-wrapped bars, slight delays, inconsistent feeding from upstream, etc. caused major problems for the robots. This affected the product spacing, which in turn affected the “instantaneous rate” at which the product would come to the cell. Bars that were ½” closer than they should be resulted in a momentary, instantaneous spike in flow to 180 bars per minute. Conversely, gaps in product flow resulted in momentary dips in product flow down to 100 bars per minute. In order for the process to work, the robots needed to be able to handle these instant spikes, and fill in the instant gaps to ensure that the right number of product went into the box every time. This variability was something that did not come up in the simulations. The variability resulted in the robots sometimes missing bars (sending them off the end) and sometimes not having enough to pack the trays (partially filled boxes).
The other major issue faced by the team was placement accuracy on the 1-pack pattern. The speed of the tray conveyor, coupled with the scan time limitations of the robots made it difficult for the robots to be consistent. The scan time of the robots was approximately 12 milliseconds, with the speed of the conveyor being 1 millisecond of time equating to approximately 1mm of travel. Therefore, the drop-position of the ice cream bar could vary by approximately 12mm. This resulted in some partial or complete misplaces by the robot. Due to the above issues, the performance of the cell was far from reaching the designed performance. Since it was already installed on the customer’s floor, this meant that the production floor workers had to endure the decrease in performance while the JMP project team worked to resolve the issues. This created a difficult environment for the project team to be successful and achieve buy-in from the workers who would be running the cell.
As it can be seen, there were several barriers to success with the project; however, this was the time when the commitment and integrity valued by JMP and embodied by the project team shone through. The team got to work solving the issues methodically while developing strong relationships with the production floor workers they worked beside every day.
In order to deal with the variability in flow, JMP replaced the entire in-feed conveyor system with a unique conveying and buffering system. This buffering consisted of a series of independent drives that allowed the product to be conveyed at different speeds on different sections of the conveyor. This conveyor would decelerate the bars, forcing them to group together and align in the process. The conveyor would then accelerate them before presenting them to the robots, which would create a uniform, consistent gap between them. This approach enabled the flow variability to be smoothed out and made consistent. This meant the robots could run at a continuous steady state rate, as opposed to dealing with instantaneous spikes or dips in flow.
The project team then set to the task of getting all pack patterns running at full rate, handling incoming bars and packing the trays fully without missing any. At this point, the mood among the production floor workers had greatly improved and many of them had become partners, making the project team aware of both issues and successes.
Though the team was enjoying more successes than setbacks by this point, the issue of drop accuracy on the 1-pack pattern threatened the project’s success. A dedicated motion controller was added to the system to specifically control the robot grippers. JMP worked with the robot vendor to tie the motion controller in with the robot software and the rest of the system. The motion controller’s scan time was approximately 100 times faster than the robot; therefore, its accuracy was dramatically improved over performance of the robot by itself. The team was confident that this was the final hurdle.
After implementing the new motion controller and eliminating the last performance issue, the system ran flawlessly and met the desired performance targets.
The real lesson learned on this project is simple – make sure that the test environment and product flow in your shop matches what will be experienced at the customer’s site. The testing at JMP’s shop for this project involved small batch runs that were consistent in flow. Since this project, JMP has implemented other above-the-belt food handling projects and has gone to extensive measures to simulate a product flow at their shop that is representative of the final system. The cost to do so is higher up-front, as you need to accurately mimic the equipment you are loading (in this case a boxer), develop a strategy to run a continuous flow that matches the plant environment, and often cool the system to avoid thawing of product (often a challenge on frozen food products when commissioning in the summer months). Despite these challenges, these steps highlight performance challenges and surprises before they get to site and make them easier to deal with and manage.
Throughout this project, the team at JMP was faced with many obstacles to success. There were many examples where the team overcame difficult situations to achieve ultimate success, both in the performance of the cell, as well as gaining the trust, respect, friendship, and buy-in of many of the customer’s frontline staff involved in the project.
Above all, JMP’s commitment and integrity was most evident in this project. JMP’s commitment was on display among the project team who put in many long hours on-site troubleshooting, reworking, designing, programming, and testing, all around the customer’s production schedule. The project team’s integrity guided them to put in everything they could to ensure that they would deliver on the promise of performance to the customer. This was then reinforced by the president who was able to assure the customer that the team would deliver on their promise or he would provide them a full refund and remove the equipment.
The feedback from the customer stands as a testament to the commitment and integrity of JMP: that never in over 30 years in the manufacturing industry had the customer worked with a vendor that had been more committed to a project nor had the integrity to stand behind their work.
Though the project had some rough moments, it serves as an exceptional example that it is never too late to do things right, and that JMP doesn’t walk away when the going gets tough.
An Interview with the Client
We interviewed the Project Manager (Engineering) at the plant about the project at its completion to get his opinions on its success.
JMP: What were the factors in your decision to choose JMP?
PM: You had very competitive pricing, as well as the excellent support and service that we were accustomed to after years of working together. You also had the ability to integrate into our existing systems with very little manipulation of our current systems.
JMP: What were you doing before we implemented the automation solution on that line:
PM: We were packing by hand into buckets. It was our top ergonomic issue in the factory. That line was running 155 bars a minute! Our employee health was the most important issue.
JMP: What was your initial experience with JMP like?
PM: Your dedication and service is first class!
JMP: What were your expectations from the project?
PM: We needed to be able to implement the robotic solution for the 600 line that removed ergonomic issues while maintaining current efficiencies.
JMP: We know there were challenges in getting the solution running where it needed to be. What did that look like from your perspective?
PM: We would have liked it to work off the hop, but your support and “doing whatever it takes” approach while investing your time, effort and resources for the long haul did make it work. We appreciate that.
JMP: What was the key factor or deciding moment that made you believe that JMP would resolve the challenge and make the solution work as promised?
PM: Given the commitment that JMP showed from all levels of the organization, I was confident that a solution would be reached. From a technical perspective, in March 2012 JMP showed that a new EOAT controller for the gripper cylinders and vacuum would greatly reduce the variability of bar placement for singles. This solution was trialed and was successful.
JMP: Are you happy with the end result? Was it worth the effort to get to this point?
PM: The end result of this project is that a major ergonomic issue has been eliminated from our factory. The safety and wellbeing of our employees is paramount; and JMP helped us to make our factory a better and safer place to work.
JMP: Would you recommend JMP and this solution (PickFlex) to others with similar material handling challenges? What would you say?
PM: Yes! You guys are certainly, as far as service and tech support partners go, completely committed to the solution and just what we were looking for.