When people think of the Tesla assembly line, the word “automation” often comes to mind. “Automating intelligently” is probably a more accurate term to describe Tesla’s automotive assembly, as a combination of artificial intelligence and automation has marked the way that Tesla manufacturing is done.
Intelligent automation systems detect and produce vast amounts of information and can automate entire processes of workflows, learning and adapting as they go. Applications range from collecting, analyzing, and making decisions, to guiding advanced robots. Automating intelligently has helped Tesla to transcend conventional performance and achieve unprecedented levels of efficiency and quality.
Think about the changes that the company’s Fremont, California, factory has undergone since it was known as New United Motor Manufacturing, Inc. (NUMMI), a joint venture between General Motors and Toyota. GM CEO Roger Smith pitched the idea that GM would gain some technology and insights into Toyota’s production system, and Toyota would get a taste of trying to apply its systems and culture on a US workforce.
That experiment failed by 2010, but today Tesla’s version of the factory there, with 5.3 million square feet of manufacturing and office space on 370 acres of land, is one of the world’s most advanced automotive plants. Tesla says it is focused on achieving the world’s most automated manufacturing systems while ensuring its large factory workforce is trained in the advanced skills unique to Tesla’s production processes. At the time that Tesla assumed control of the Fremont plant, Gilbert Passin, the Tesla manufacturing chief at the time, said:
“Why wouldn’t we try to integrate as much as we can here? The opportunities are huge. We have stamping in-house. We have plastic injection molding in-house. So, to leverage the equipment we have is key from a financial standpoint, rather than pay a lot of logistics, packaging, and handling costs. Plus, the ability to control our own destiny, our own quality, in our working process, is huge.”
The Fremont facility, which is pioneering vertically integrated manufacturing, does control a good deal of the Tesla destiny. Tesla is largely self-contained — a significant portion of the Tesla supply chain is company-owned owned, and each division within the company supply chain produces a different product. The products are then combined to satisfy a common need.
Musk’s approach to automating intelligently is derived from his belief in first principles thinking, which means starting free from any pre-existing ideas, yet grounding conceptual frameworks with core fundamental basics. It’s an approach that leaves one open to problem-solving and creating innovative ideas, and, in Musk’s case, has led to unrivaled automation innovations.
Understanding the Future Sometimes Means Ignoring the Skeptics
Simply believing that an all-electric car could be viable through battery power seemed naive to many people at one point. Yet Musk challenged that notion — looking ahead with a focus on automating intelligently.
“Someone could — and people do — say battery packs are really expensive and that’s just the way they will always be because that’s the way they have been in the past. They would say, “It’s going to cost $600/kilowatt-hour. It’s not going to be much better than that in the future.”
As we already know, Tesla is far past that point.
Musk has vigorously pursued automation solutions to produce all-electric cars. He recognizes that automating intelligently to smooth cost and efficiency requires new approaches and a constantly revised, forward-looking vision of an automated future. Harvard Business Review reasons that the factory of the future “will look different because we will have invented entirely new processes and designs for building cars requiring entirely new manufacturing techniques.”
Robot process automation combines artificial intelligence, according to Larent and co-authors, including integrating natural language processing, machine learning, autonomics, and machine processing into automation. Advances in machine learning techniques, improvements in sensors, and ever-greater computing power have helped to create a new generation of hardware and software robots with practical applications in many industries, with Tesla at the forefront.
“Our internal code name for the factory, the machine that builds the machine, is the alien dreadnought,” Musk said during a 2016 investors’ call. “[When] our factory looks like an alien dreadnought, then we know it’s probably right.”
The Tesla Factory Today & The Future To Which It Points
In a visit to the Fremont “disruption” factory, Wired observed several innovative automation processes. Below are three.
The chassis was welded together by Cold Metal Transfer welding robots designed to work on Tesla’s aluminium frames. The metal was chosen over steel because it’s lighter and also safer, as it absorbs more energy in a collision. Cold Metal Transfer technology has revolutionized the welding of dissimilar metals and thicker materials by producing improved weld bead aesthetics with controlled metal deposition and low heat input.
Bernstein analysts Max Warburton and Toni Sacconaghi describe how huge numbers of Kuka robots work on the assembly line — not only automated stamping, painting, and welding, but also automating final assembly.
The Fremont factory’s giant Schuler SMG hydraulic press is 7 stories tall and is the largest stamping press in North America. 50 trucks and 30 rail cars transported it from Detroit to Fremont. Tesla bought the $50 million machine for $6 million to turn giant rolls of steel into Model 3 body parts. There are only 35 such presses in auto manufacturing worldwide, and the press produces unique parts like the Model 3 front fender, which engineers say has greater depth in a single piece of stamped steel than any other fender in production.
[Note: CleanTechnica’s own Matt Pressman toured the Fremont factory in 2016 — check out his and CNET’s great narrative of the factory experience here.]
Tesla Paint Shop as an Example of Automating Intelligently
Like with most automakers, the Tesla paint shop is nearly fully automated, yet it is still one of the most expensive and space-intensive sections of the factory. Robots, instead of humans, perform most tasks — applying protective corrosion coats, sealant, primer, basecoat, and clear coat to achieve the highly polished finishes. In May 2018, Musk said, “General assembly is probably our biggest risk, and I’m refocusing personally on that a lot in the next — in the coming month. And then our paint shop is maybe the second biggest risk after general assembly.”
Automating intelligently for paint application efficiency and cost will require Tesla to develop new processes. The Harvard Business Review authors predict that an experimental approach of applying a single film over the car and then baking it on, like in a pottery kiln, may be the next step forward for automotive painting. They also point to 3D printing of the entire car body in the color a customer orders, completely eliminating the need for a traditional paint shop and body shop. “Whatever it is, it will have to be more than adding a few more robots into the mix to make a significant difference in the cost of producing an auto,” they argue.
The factory of the future is closer than we think. Recognizing that a strong economic impetus exists to combine purpose-built, safe, and precise products, Tesla is creating ways to manufacture environmentally neutral products with far less energy and material throughout the value chain. New integrated technologies, factories that operate in networks, digitizing everything, connectivity, making data a key asset in the manufacturing process, new organizational and architectural infrastructures, and AI — the factory of the future is the Tesla factory of today. And, you know, it’s an amazing manufacturing process to see.
[Note: An email allegedly sent by the Tesla CEO at 11:57 pm Pacific Time on Sunday, June 17, 2018, claims that an employee modified code of the Tesla Manufacturing Operating System and sent sensitive internal data to external parties without authorization. If this report turns out to be valid, it would point out how valuable the Tesla Manufacturing Operating System, a set of basic commands for Tesla’s manufacturing lines, currently is to competitors within and outside the auto industry. Stay tuned.]