Robots Driving Automotive – From Early Failures to Success
The field of AI-powered humanoid robots is currently experiencing a development surge. For example, BMW Group is currently testing how they can safely use humanoid robots in automobile production. Maybe they should talk to the folks at GM and Tesla.
During a trial run at the Spartanburg BMW Group Plant, humanoid robot, Figure 02, successfully inserted sheet metal parts into specific fixtures that were eventually assembled as part of the chassis.
The trial resulted in an understanding of the requirements to integrate multi-purpose robots into existing production systems safely. These requirements include how the robots communicate with the system under real conditions. After the trials, the experts at BMW Group and Figure are collaborating to prepare robots for future applications in production based on the findings.
According to California-based Figure, Figure 02 is the world’s most advanced humanoid robot currently available. It features two-legged mobility and advanced agility, making it valuable for physically demanding, unsafe, or repetitive processes. Figure 02 has 3x the processing power of its predecessor, improved voice communication, cameras, microphones, and sensors, a high-performance battery, and fourth-generation human-scale hands with 16 degrees of freedom and human-equivalent strength.
The robot is capable of fully autonomous execution of human-like and two-handed tasks needing complex grasping and coordination of both hands in unison. It can place various complex parts with millimeter accuracy and can walk dynamically.
Currently, there are no Figure AI robots at BMW Group Plant Spartanburg, and there is no definite timetable to bring them on, but estimates are in the next 12-24 months. BMW Group will use them in the body shop for sheet metal work and warehouse operations. It will continue to work with Figure for data capturing and training capabilities for the Figure 02 robots.
Other companies are developing similar robots. Apptronik, for example, is in a partnership with Mercedes with the Apollo robot, and Boston Dynamics continues to develop its Atlas robot for applications, including auto-manufacturing tasks. With its vast car production facilities, Tesla is also in the race with its Optimus robot. Each company is taking a slightly different approach.
It wasn’t always as smooth sailing as the BMW/Future trials. For example, under the leadership of Roger Smith in the 1980s, General Motors set out to replace people with machines to counter Japan’s gaining ground and return GM to the dominance it once held in the global auto industry. GM’s plant productivity, which had lagged Toyota’s for years, declined further from 1984 to 1991, a failure of GM’s automation push. Some plants ran at 50% capacity because of glitches in computer-integrated systems, and two major strikes in the U.S. and Canada in the mid-1980s occurred based on labor relations regarding the changes. GM’s share of U.S. auto sales fell to 41% in 1986.
Rather than adopt Toyota’s lean manufacturing techniques, Smith’s obsession with robotics took over, and GM missed the essence of Toyota’s low-cost production success. Industry expert Robert Lutz, who witnessed changes in the auto industry over the years as a senior executive at GM, Chrysler, and most recently, Ford, explained, “The thought was if we can do a fully automated factory and get rid of all the labor, we would have plants that run day and night fully automatically. But with these totally automated facilities you lose all flexibility, and they are extremely capital intensive. They were prisoners of the great North American manufacturing cost accounting system that says, as you eliminate labor, your costs go down. But what they forgot was they were getting rid of direct labor but replacing it with indirect labor and huge capital costs. Technicians and other people needed in an automated plant were much more expensive than the hourly laborer…”
For Smith, robotics represented the Holy Grail, which would simultaneously solve all of GM’s problems. GM invested $90 BILLION over ten years in this quixotic (if not foolhardy) quest. The robots famously painted each other instead of painting cars. In the book Comeback, by Paul Ingrassia and Joseph B. White, “…as Hamtramck’s assembly line tried to gain speed, the computer-guided dolly wandered off course. The spray-painting robots began spraying each other instead of the cars, causing GM to truck the cars across town to a fifty-seven-year-old Cadillac plant for repainting.” The robots also could not tell one model car from another. They installed Buick Riviera bumpers on Cadillac Sevilles.
In 2017, Elon Musk attempted a lights-out factory to mass-produce Tesla’s Model 3. Tesla rapidly experienced production delays, and, in Musk’s own words, they struggled to navigate a “crazy, complex network of conveyor belts.” Tesla also reversed course, abandoning some of its investments in automation and scaling up its skilled workforce. “Humans are underrated,” Musk said.
Robots are coming back to the production floor without the barriers experienced by the industry so far. In reality, productivity and flexibility still require humans and will for the foreseeable future.
