A recent McKinsey and Co. report entitled “Where machines could replace humans and where they can’t (yet)” looked across industry sectors to determine which jobs could be replaced by machines based on currently demonstrated technologies:
Almost one-fifth of the time spent in US workplaces involves performing physical activities or operating machinery in a predictable environment: workers carry out specific actions in well-known settings where changes are relatively easy to anticipate. Through the adaptation and adoption of currently available technologies, we estimate the technical feasibility of automating such activities at 78 percent.
If you view this report through the lens of warehousing, many physical activities performed in a warehouse throughout the workday are predictable, making many warehousing jobs highly susceptible to automation.
The future is coming.
Although many warehouse tasks are predictable, many still require flexibility to handle the wide variety of products found in most warehouses from small unit picks to irregularly shaped products. The variability in different sized products and packaging is perhaps the largest barrier today to making robots ubiquitous in warehouses.
Currently most warehouse robots are designed with a single, or small number of related, package sizes and shapes in mind — or the merchandise is prepackaged or palletized for easy automated handling. In Amazon’s case, where Kiva robots are used extensively, human workers still do the actual picking of pieces from shelves as the robots have trouble picking products of unpredictable sizes. The next generation of robots, however, will incorporate advances in deep learning - the algorithms that powered Google's AlphaGo to defeat the world champion Go Player Lee Sedol - to identify products, pattern-match the size and weight and then identify the most optimal strategy to pick, pack, and ship. This technology is already available and has been used in facial recognition technology and for self-driving vehicles.
In the short-term, the technology continues to improve and is expanded to warehouses that manage multiple different types of products. Human workers and robots work side-by-side and robots perform 50-60% of the work. In the medium to long-term, advances in deep learning and artificial intelligence enable robots to perform 90% - 100% of the work depending on the type of warehouse. Robots have enough dexterity and built-in intelligence to allow human workers to disappear completely from certain types of warehouses. In semi-automated facilities, a handful of human workers would be required to manage exceptions and deal with oddball-sized products. In fully automated facilities, human workers in centralized control centers could manage the robots across multiple facilities.
It is important to note that the Mckinsey study only considered currently available technology so as the technology improves via Moore’s law, the economics will make more sense. Ultimately, the business opportunity is too great for companies not to aggressively pursue robots in warehouses. From a purely bottom-line perspective, robots don’t require health insurance, vacation, or sick leave and can work 24 hours a day. They don’t need specialized training. They don’t have to adjust to seasonal changes in product lines. The future is not quite now but it's rapidly approaching and it will have dramatic implications for the