ROI Case File No.589: They Tried to Move Everything onto Machines Without Choosing Which Motions Were Needed

They Tried to Move Everything onto Machines Without Choosing Which Motions Were Needed

---

Chapter 1: They Want a Robot, but the Specification Won't Firm Up

"We want to bring humanoid robots into our film manufacturing plant."

Kaname Shoda, engineering department manager at NeuroDynamics, said this as he laid out the situation. "It's a plant handling paper and film. See with vision, hold with two arms, move up and down, adjust force with touch. We want to move what people are doing onto a robot exactly as-is. We're looking for a company with the technical expertise."

"Which tasks specifically?" Claude asked.

"Peeling and picking sheets one at a time, transport that requires delicate force control, and visual inspection," Shoda answered. "All of them rely on human senses. Then there's transport in the logistics process. That one is high-frequency and hard on the workers' bodies. I was hoping to hand all of it to robots at once."

"Among those, have you separated the motions worth a human doing from the ones that aren't?" I confirmed.

"...We haven't," Shoda answered. "I was working on the premise of moving everything people do onto machines as-is. Which motions carry value and which shouldn't exist in the first place—we've never sorted them."

"Move everything across and you end up reproducing the unnecessary motions in machinery too," I replied. "Let's break this down with LEAN."

Chapter 2: LEAN Asks—Keep Only the Motions That Create Value

"This case calls for LEAN."

Claude wrote "Value / Flow / Waste" on the whiteboard.

"LEAN—lean production—is a framework for reviewing processes against what the customer values and removing the motions that create no value," I explained. "The essence is not automating the current work as-is. Human motion mixes value-creating movement with movement like transport and waiting that shouldn't exist at all. Move it onto a machine with the waste still inside, and you've built expensive waste. Sort it first."

"First, let's measure the current cost," Gemini said, opening ROI Polygraph. The data Shoda provided went in.

"The monthly costs are out," Gemini read off. "Visual inspection and single-sheet picking averages 190 hours per month; at ¥3,500 per hour, that's ¥665,000 per month. Physical strain and turnover from high-frequency transport averages ¥420,000 per month. Defect escapes from inconsistent inspection quality average ¥380,000 per month. Night-shift and support-shift costs from reliance on manual labor average ¥340,000 per month. Development spinning its wheels because the requirements won't firm up averages ¥320,000 per month. The total is ¥2,125,000 per month. Annualized, roughly ¥25,500,000."

Shoda stared at the figures. "I was only looking at the burden of the work. Once you add the defect escapes and the development costs sitting idle while the specification won't settle, it comes to this much."

"Then let's design with LEAN," I continued.

---

[Value—Name the motions that create value for the customer]

"First, value," Claude said. "Picking film one sheet at a time reliably, handling it with a force that won't damage it, and finding defects. These three directly create product quality. This is where human senses are doing real work. So if you replace them, treat them as territory where accuracy cannot drop."

---

[Flow—Connect the processes and see where things stop]

"Next, flow," Gemini continued. "Line up the processes and see where the material stops. Queuing for inspection, waiting for transport. Time spent stopped creates value for no one. This is the first place to act."

---

[Waste—Strip out the motions that don't need a human first]

"After flow, the waste," I continued. "The high-frequency transport in the logistics process is frequent, simple in motion, and creates no value. It doesn't need human senses either. Automated guided transport is enough here—no two arms, no tactile sensing required. Strip out the waste alone and most of the physical strain and turnover disappears."

---

[Small—Build the necessary functions in order]

"Last, going small," Claude continued. "Don't build the full-body type all at once. Ship transport automation first, then vision and two-arm single-sheet picking, then tactile force control. The specification won't firm up because you're trying to define all of it at once. Separate the functions and you can decide them one by one."

---

[Calculating the payback]

"Let's run the numbers with ROI Proposal Generator," Gemini proposed.

• Initial cost: Process value analysis, deployment of automated transport equipment, development of a vision-and-two-arm single-sheet picking mechanism, inspection support implementation, and requirements documentation — ¥5,200,000 total
• Monthly cost: Equipment maintenance plus ongoing system operation — ¥240,000 per month
• Monthly savings: Automation of the transport process = ¥520,000 per month (assuming a 70% reduction), reduced defect escapes through consistent inspection quality = ¥360,000 per month, suppression of physical strain and turnover = ¥300,000 per month, reduced night-shift and support-shift costs = ¥280,000 per month — ¥1,460,000 per month in total
• Net monthly savings: ¥1,460,000 − ¥240,000 = ¥1,220,000 per month
• Payback period: ¥5,200,000 ÷ ¥1,220,000 = approximately 4.3 months

"A payback of just over four months," Gemini summarized. "What makes it work is not moving human work across wholesale, but choosing only the motions that create value. Try to reproduce everything with a humanoid and development drags on while simple transport gets handed to expensive machinery. Strip the waste first and it pays off quickly. The investment doesn't swing at air."

Shoda looked over the figures. "I thought reproducing human motion as-is would do it. Without choosing the needed motions, you turn the waste into machinery too."

"LEAN is a tool for keeping only the motions that create value," I replied.

Chapter 3: An Implementation Plan That Chooses the Value-Creating Motions

"Let me lay out how to proceed," I said, standing at the whiteboard.

"Month one—map the processes and sort value-creating motion from waste. Month two—measure the stoppages and waiting, and fix the requirements. Months three and four—automate transport in the logistics process. Month five—prototype and validate the vision-and-two-arm single-sheet picking mechanism. Month six—implement inspection support and verify results. Month seven onward—add tactile force control and expand to more processes."

"If we're building anyway, wouldn't a robot that can do every motion from the start be better?" Shoda asked.

"While you wait for the one that can do everything, nothing on the floor changes," Claude replied. "Transport automation alone can be in place in a few months with existing technology. Reduce physical strain and turnover that way, and that result funds the next development. Tactile control is difficult and takes time. Splitting them into an order of effectiveness, rather than bundling them into one machine, comes before building it all at once."

Shoda took notes. "Before moving it onto machines, choose the motions that are needed. I can see the order now."

Chapter 4: The Day They Chose the Needed Motions and the Floor Got Lighter

Ten months later, a report arrived from Shoda.

The logistics process changed first, with transport automation. "Transport runs that used to be dozens of round trips a day now happen unmanned. Nobody quits from a wrecked back anymore," Shoda wrote.

Single-sheet picking took shape as well. A mechanism narrowed to vision and two arms stabilized film separation. "We built only the necessary functions instead of waiting for the full-body type. As a result, it reached the floor faster than I expected," the report said.

The largest change showed up in their thinking about automation. From trying to move human motion across wholesale, they moved to choosing the motions that create value. "I was writing specifications on 'make it do everything a human does.' Once we kept only the value-creating motions, what we had to build was cut in half," Shoda wrote.

Inspection quality stabilized too. The variance from relying on human eyes was smoothed by the support mechanism. "Missed defects dropped, and escapes stopped. The waves from inspector fatigue are gone," the report said.

As a secondary effect, their view of the process changed. Distinguishing being in motion from creating value took root. "We stopped equating busy movement with work. Now we ask whether this motion creates value," Shoda wrote.

At the end of Shoda's report, he had written: "I thought the factory automation problem was that no technology could reproduce human senses. But the real problem was that we tried to move everything onto machines without choosing which motions were needed. The moment LEAN sorted value from waste, what we should build was decided. Before moving it onto machines, choosing the needed motions came first."

The day a company that had tried to move everything onto machines became a company that chooses the motions that create value, factory automation had changed from wholesale reproduction of human work into a design that keeps only value-creating motion and strips the waste away, the report noted.

"Factory automation consultations almost always arrive in the form of 'we want a robot that can do what a person does.' But before you move it onto machines, there is something to ask. Does that motion create value? What LEAN asks about is value, flow, and waste. Sort them and the machine you must build gets smaller and pays off sooner. The day a company that had tried to move everything onto machines managed to choose the needed motions, what changed was not the robot's performance but the very perspective of keeping only the motions that create value."

---

Related Files

Tools Used

• ROI Polygraph — Visualizing visual inspection and single-sheet picking hours, physical strain from transport, and defect escapes
• ROI Proposal Generator — Investment-recovery simulation for factory automation and robot deployment driven by sorting out the value-creating motions