What does it take to get to Lights Out Manufacturing?
Interview with Bernard Casse, CEO of RIOS Corporation
Everyone is all abuzz about Industry 4.0. In recent months we’ve seen acceleration in the deployment of AI, automation and robotics across foundational industries as the US seeks to emerge from the pandemic, increase GDP growth, and offset extreme labor shortages in skilled jobs across every major sector of the economy.
In your recent Medium article, you said that there are 345 million people on the factory floor (13 million in the US) and less than 2 million industrial robots. With an estimate of only 250,000 industrial robots being deployed annually and each robot displacing 5.6 jobs, that’s 1.5 M jobs lost to robots globally. Of course not all jobs can be automated so easily and many of the early deployments are more likely to augment humans.
I’d like to explore this topic of “Lights out Manufacturing” with you. With an increasing global population, pandemic risk and increasing fragility of our supply chains, we have an even greater need to create factories that can operate without humans.
Q: Do you believe that we will see 1.5M jobs lost to robots in 2021?
This is a thought-provoking question — a lot of people’s natural reaction to this would be “Oh my God, robots are taking over people’s jobs and we’re all doomed!” But there is more to this common misconception that I’d like to dispel before directly answering the question. The fundamental premise is that automation technologies underpinned by artificial intelligence and robotics will dramatically spur productivity and economic growth. Historically, technology-driven innovations yielding significant economic growth led to elimination of certain roles but, interestingly, contributed to more jobs. If we look at the automotive industry (the largest adopter of robots) from 2003–2008, we’ll realize that employment rose to 22% from 824,400 to 1,005,000 jobs after a 65% increase in industrial robots. Jobs were lost by automation, but new occupations that did not previously exist were created. We expect the same timeless pattern to unfold.
The extent to which robots displace workers in 2021and beyond will depend on the pace of technology development, economic growth, and importantly the rate of adoption. What we know for sure is that robots will displace certain roles and people will need to switch occupational categories and upskill. Though, as history taught us, these new jobs will be higher-level, be better paying (and in some cases will reduce the average hours worked per week), and allow people to enjoy more leisure time.
Q: What types of industries will deploy robots the most aggressively and what types of jobs will be lost next?
We’ve observed that industries that are impacted in a “positive way” by the pandemic are more enthusiastic in implementing automation. Those pandemic-proof industries include healthcare, manufacturing, food services, and agriculture and they are seeing an uptick in customer demand. We’re already seeing cleaning robots and telepresence robots being deployed in hospitals. Tele-operated robots that can back up medical workers on the front lines are being trialed. Security robots are being rolled out. In the food services sector, delivery robots are on the rise. Also, when it comes to food, manufacturers are fast-tracking contactless ways of preparing and handling food. Notice that there’s a trend here — people want to limit human interactions as much as possible since “hygiene” is becoming a major concern. The adoption of automation (and consequently the displacement of workers) in these critical sectors is now driven both by chronic labor shortage and the new social distancing directives.
In the manufacturing sector, there’s an additional factor that comes into play, which is the push for reshoring — we’ve (as a country) suddenly realized that our supply chain network, which relies on foreign countries, is very fragile. Today, there is a strong desire by US manufacturers and other stakeholders to bring manufacturing back and make it more resilient by leveraging (and investing on) robotics and AI. By capitalizing on these bleeding-edge technologies, we’ll be able to maintain global competitiveness. In manufacturing, the roles that will first evolve are low-skilled jobs in which humans are doing repetitive tasks in controlled environments (i.e., pick and place, kitting, machine tending, etc.). We’ve seen factories starting to deploy cobots to tackle those scenarios. In agriculture, adoption will be slow but we anticipate a massive shift in the next 5–10 years — we’ll see fruit picking robots to fully-automated vertical farming solutions emerge.
Q: You say that 30% of tasks can be automated by traditional robots, but the remainder of the tasks are going to be elusive. What are some of the hurdles to deploying more industrial robots?
There are only two main hurdles: technology and adoption. From a technology standpoint, traditional robots are not there yet to automate the remaining tasks. The way traditional robots work is that they’re scripted (i.e., they literally have to be told every move) to do a task and they can’t deviate from that script. This also implies that they need to operate in tightly controlled environments. When tasks in factories involve multi-SKUs (i.e., many objects with different shapes and sizes) and unstructured or even semi-structures environments, these robots fail miserably. In the first place, it is nightmarish to even program them to do these kinds of tasks, and every slight fluctuation in the environment (e.g., moving a screw by an inch) will simply break them. That’s why RIOS is building dexterous, AI-powered robots (robots 2.0) — we’re building futuristic machines that possess more dexterity, higher cognitive skills and autonomy than traditional robots. Programming robots 2.0 by “drag-and-dropping” building blocks to rapidly solve complex tasks is a quantum leap forward. We’re also building the kind of intelligent machines that are much more robust to the dynamics of the environment. In a nutshell, we’re building both capabilities and speed, which will quell the hurdle on the technical front.
The other hurdle that innovators or entrepreneurs have little control over is the rate of adoption. Not all companies readily embrace leading-edge technologies — whenever innovations are introduced to the world, only early adopters go for them. The early adopters are willing to take risks — they are typically seeking a fundamental breakthrough in a core business problem or strategic opportunity. It’s a group that can start to see how these innovations can help them achieve their own vision and organizational goals. Most businesses are laggards — they’re risk-averse and they require demonstrated, standardized and proven solutions before adopting anything. So, that’s a showstopper in rapidly deploying robots — but that dissipates over time as the technology matures and when there’s more widespread adoption.
Q: What are some of the some of the promising technologies to address the technical hurdles?
Today, there’s a confluence of technologies that allows us to build the robots of the future. We can leverage a lot of infrastructure that was built over time — it’s a similar scenario to the year 1995 when Amazon capitalized on the Internet to start building its juggernaut business. In 2020, we’re riding on four pillars: advancement in hardware/electronics, computing horsepower (GPUs + edge computing), AI, and cloud computing. We couldn’t have built these types of dexterous and intelligent machines 5 years ago — at the very least not at the unprecedented speed that we’re developing them today, and not with the current level of performance. But at a high level, in hardware, there are new ASIC chips, embedded processors, and novel sensors that didn’t exist a couple of years back. On the software end, we now have the robot operating system (ROS) platform and powerful techniques like deep learning and reinforcement learning. We’ll see more companies emerging in the next few years that will piggyback on this massive infrastructure to address technical hurdles from several dimensions. There’ll be a convergence of technologies at some point, and there’ll be more than one approach to address all the technical challenges, all by leveraging these four pillars.
