eCommerce revenues are up considerably, topping $390 billion last year and doubling those of 2011, increasing the pressure on retailers and warehouses to pack more orders faster while maintaining high accuracy rates.
Right now, that job is one entrusted to a group of 950,000 people in the United States alone; 262,000 of those jobs were added within the last five years alone. According to the Labor Department, however, worker shortages are becoming problematic, especially during peak seasons.
Handling Worker Shortages in the Warehouse
As costs climb and warehouses compete for the limited picker pool, there are companies working on the issue from another angle.
Using advanced technology and machine learning, robots are being taught to identify items more like a human picker would. In the past, robot warehouse pickers have failed to impress because they had to be programmed with information on each individual item in a warehouse, which could be very time-consuming, especially when the warehouse rotated stock frequently.
The learning robots, though, could change the entire game. Experts believe that swapping human pickers for robots could shave 20 percent off the labor cost of fulfillment. In addition, robotics companies are touting robots that can move products 50 percent faster than humans. It’s the perfect marriage of cost-savings and a much-needed boost in pick speed for the increasingly demanding eCommerce machine.
Research Sharing to Speed Up Learning
The biggest challenge for warehouse robot pickers is creating the incredible databases needed to teach robots how to grip differently shaped items.
Essentially, the picker robots are taught how to manipulate different 3-D objects, one at a time in a virtual environment. Each pick, whether successful or not, is recorded so the machine will understand how to do its job better.
Programming the databases requires an immense number of human hours, which is why companies like Amazon and Siemens AG are working together to fund a team at UC Berkeley. Their goal is to build an open-source object database that can be used in any type of automation system. Currently, they’re simulating millions of attempts at picking 10,000 different items, but the hope is to expand that number dramatically.
Robot pickers won’t replace warehouse workers anytime soon, they’re currently at least a year away from commercial availability. With the rapid growth of eCommerce order fulfillment, it’s unlikely that robot pickers will push human pickers out of work anyway, but they may be able to speed up shipment times and increase accuracy for warehouses across the country.