Comparing Automation in Material Handling and Manufacturing
By Meena Vembusubramanian, Senior Director of Product Management, Veo Robotics
I started my career in the factory automation space at Rockwell Automation and then moved to Kiva Systems (now Amazon Robotics). Having spent a fair amount of time thinking about automation in a variety of contexts, including factories and warehouses, I’ve gained a broad perspective on the transformative potential of automation across both material handling and manufacturing applications. The recurring theme has been identifying patterns that enable people to do what people are good at—judgment, decision-making, and fine motor control—and leveraging machines for what they are good at—carrying heavy loads, repeating tasks, and traversing distances. Veo Robotics, where our goal is to help customers design the next generation of factories, is bringing me full circle in creatively applying this insight to manufacturing.
Unlike material handling, manufacturing does not have a very straightforward formula for applying automation that we can replicate broadly—but this is exactly why I am particularly excited about Veo’s technology and the potential it holds for manufacturing. Manufacturing processes are specialized and varied, each presenting a range of interesting and challenging problems to be solved, and the people who build factories are creative engineers who have been innovating on how things are made since long before robots even existed. Moreover, cracking this nut will have a massive impact: the durable goods manufacturing market is about ten times the size of the warehousing market. Even with recent high growth rates in material handling, its upper bound is still well below the current size of manufacturing. 1
Putting things in perspective…
To understand why manufacturing automation is at such an exciting point in time right now, it’s helpful to first consider the success of automation in material handling. Many people, when they think about robotics, think about automatic guided vehicles roaming a warehouse. This is likely thanks to the amount of media attention Amazon’s acquisition of Kiva Systems received in 2012, due to its strategic significance and size. Since then, the material handling industry has seen a wide range of funding activity. This is in large part thanks to recent changes in consumer behavior that are driving the need for better, faster, cheaper material handling—consumers want a lot of choices, they want to make purchases from the comfort of their homes, and they want to receive everything they’ve ordered in two days—and then be able to return half the items for free. The big players in retail are willing and able to meet these types of demands and to do so on an increasingly global scale. But smaller specialized retailers can have a harder time finding a way to make the logistics work without breaking the bank.
We are well past the point where throwing people at the problem is a viable solution, even for small operations. In a traditional low- to no-automation context, a person will hit 40–60 picks per hour on a good day. Meanwhile, I have seen various high- to full-automation or blended (human + automation) approaches readily achieve over 200 picks per hour. This is exacerbated by another trend I have consistently heard about from customers: it has been becoming increasingly hard to hire, train, and retain people for industrial jobs across multiple geographies.
The good news is that there is a relatively well-understood, generalizable approach to automation that can be applied broadly in warehouses: replacing walking. Kiva’s solution was successful because it focused on and tackled this specific part of the fulfillment process where automation could deliver the most value. With Kiva, there was a clear way for people to focus on what they were good at (differentiating between and grasping multiple objects, etc.) while robots took care of the rest (see: The Magic Shelf—still one of my favorite pieces of automation marketing). This distribution of work allowed for a system that was more productive than the sum of its parts.
While there is a lot of mileage still to be gained from automating walking in warehouses, much of the industry’s innovation since then has been incremental, and multiple players continue to craft products around variations of this solution. One recurring theme is enabling a modular approach: while the Kiva system required a complete overhaul of a warehouse, new players such as 6River and Locus Robotics are able to offer more flexibility for their customers through modular solutions that augment existing processes and work out of the box with existing inventory management practices. Retailers can then scale and adapt as their business needs change. Over time, these providers will be able to improve the efficiency of existing processes incrementally through operational improvements such as grouping like orders, storing products that are often purchased together in the same locations, integrating forecasting and demand planning, and etc.
As robotics technology advances, companies are tackling end-effectors and gripping technology to automate the pick-and-place aspects of the workflow, as well. We saw this with Right Hand Robotics’ collaboration with Locus Robotics, a combination of technologies well-positioned to tackle singulation steps in a warehouse, for example. While there is certainly value on the table here, the appetite for investment in this space is limited since logistics operations are largely viewed as cost centers for an organization. All goods pass through distribution centers unprocessed, handled without value added by humans and machines. Manufacturing operations, on the other hand, consist of dozens of value-add steps to transform raw material into finished product.
How does manufacturing compare?
The same trends that have been driving growth in material handling also hold true for manufacturing as consumers’ desires for increased choice and personalization continue to make “durable” goods less durable and more disposable. This is particularly true in the electronics space, where people are accustomed to faster upgrade cycles and a broad set of choices, and like to purchase the newest device or appliance before the estimated lifetime of their current one expires. Because they are buying these products more frequently and using them for less time, the amount consumers are willing to pay on a per-device basis has rapidly fallen, placing pressure on OEMs to increase the flexibility (and therefore complexity) of their manufacturing operations, as well as to reduce costs.
The difference lies in the scale of impact these pressures are going to have in factories as compared to warehouses: the number of steps to be completed in a given distribution center tends to be somewhere on the order of 20–40, depending on how the warehouse fits into the larger operations picture (unloading, receiving, depalletizing, sorting, pick/pack, re-palletizing, labeling, shipping, and etc.). In comparison, the number of engineering steps that happen in a manufacturing facility that produces automotive components, for example, is easily on the order of 500 or more. The following figures highlight major processes in each context, and the different types of equipment required in these steps.
One challenge that stems from the complex and bespoke nature of each manufacturing process is that there isn’t a clean, generalizable application in factories akin to replacing walking in a warehouse. Compared to designing a new fulfillment operation, which often involves a well-understood blueprint and straightforward throughput and cost goals, designing a new production line allows manufacturing engineers many more degrees of freedom and many more tools with which to create a system. Each process step and workflow tends to be specific to the products and sub-assemblies that are being built. The good news is that a sophisticated ecosystem of specialized sub-assembly and component manufacturers as well as systems integrators has nucleated and grown around manufacturing centers to cater to the equipment and automation needs of factories—dating all the way back to the first lines at River Rouge.
Systems integrators are a unique and central element of this network: they are responsible for incorporating automation and controls equipment from a variety of suppliers as they build out factories and workcells. For an OEM, a systems integrator offers the benefit of integrating technology from scores of robot, controls, and automation suppliers into fully built-out and calibrated workcells capable of executing process steps to specification—all delivered and installed on site. This one slice of the manufacturing ecosystem alone reports revenues upwards of $27 billion dollars a year—equal to almost 15% of the whole material handling market.
Manufacturing engineers are well accustomed to leveraging this ecosystem to design complex, repeatable workflows. But they run into one common problem: existing options for automation limit the amount of flexibility a production line is able to sustain, making it challenging to keep up with consumer demand for variety. Adding multiple tool and tool-change options to existing workflows to handle part diversity results in skyrocketing costs, sometimes overshadowing the cost of the process step as a whole! As Alberto has discussed, in a lot of cases it’s simply not economical to invest in current automation technologies because they result in brittle, inflexible systems.
Multiple signs point toward the fact that enabling flexible production lines is one of the most pressing challenges to be solved in manufacturing. Achieving flexibility in production lines can be facilitated by allowing humans and robots to collaborate in tasks. This is precisely why we approach automation in the way that we do at Veo. We believe that industrial robots are beautiful, well-designed machines tailored to the factories they are built for, and that factories must successfully and efficiently leverage the skills of both robots and people. Veo’s approach allows our customers to keep people in the loop, working alongside powerful industrial robots—safely—doing truly valuable work that requires intelligence and dexterity while robots take care of the rest. We’re seeing the possibilities this solution enables play out with our customers today, and I am excited to see the next phase of innovation that will come from putting our technology in the hands of manufacturing engineers as they build Veo-enabled workcells and factories.
1 For context, durable goods manufacturing in the U.S. is valued at over $2 trillion, as compared to material handling, which clocks in at just under $200 billion.