Arms Race: The State of Robotic Piece-Picking Solutions

Of all the tasks performed in an e-commerce warehouse, picking is the most crucial. It’s also among the most costly. The amount of time that workers spend walking from location to location throughout the warehouse picking items for customers’ orders accounts for about 50% of total pick time and represents more than half of all operational costs. Additionally, finding and retaining staff in today’s tight labor market is difficult, often resulting in a repetitive and expensive cycle of hiring and training temporary workers, only to have them call in sick, fail to show up, or just quit a few weeks or months down the road.

It’s no surprise then that picking is the focus of many companies—including inVia Robotics—that are looking to automate part or all of the process. Automated systems come in several configurations, from large, fixed automated storage and retrieval systems, to collaborative robots that accompany workers as they walk down the aisles, to autonomous mobile robots (AMRs) that deliver totes to a packing station or virtual put wall. That movement of totes by AMRs is what robotics researchers call “gross manipulation.”

The logical evolution of robotic picking is a system capable of “fine manipulation.” That’s when a robot can pick individual pieces from a bin with an arm and a gripper. Roboticists consider fine manipulation to be the next big breakthrough, with the goal of building a robot that uses machine vision and AI to open a jar of peanut butter or pick up a wine glass without breaking it seemingly within reach. 

Robots capable of fine manipulation used in the picking process would, in theory, result in a fully autonomous system that could correctly pick and pack a wide variety of items, regardless of shape, size, weight or packaging design.

But that’s where things get exponentially complicated.

Get a Grip

There’s a well-known paradox in the field of AI and robotics that refers to the discovery by researchers that, contrary to what you’d expect, high-level reasoning requires relatively little computational power, while low-level, sensorimotor skills require an enormous amount of computational power. In other words, what’s hard for a person—like winning a game of chess—is easy for a machine. And what’s easy for a person—like picking a pen out of a drawer full of pencils—is quite difficult for a machine. 

A handful of companies working on piece-picking systems have demonstrated a measure of success in picking individual items out of totes, but there are numerous issues in deploying such systems at scale.

One hurdle to overcome is the fact that most robotic arm systems are stationary, so in order to pick items for orders, totes would need to be brought to arms by shuttle, conveyor or by other robots. This almost certainly would require a warehouse redesign or construction of a new facility.

Another issue is that the current generation of robotic arms at work in factories perform specialized tasks with grippers designed to grasp specific objects. Designing a gripper that can mimic the sensitivity and versatility of a human hand requires immense computational power, so most robotic picking arms use suction grip items, which often limits what they can lift. 

Think about the items you can order online: a box of laundry detergent, a jar of pickles, and a bag of rice. You could pick any of these items off a shelf easily with one hand despite the variations in size, shape, and weight. But for a robotic arm, it’s a different story. The box of detergent is a heavy cube, while the pickles are in a cylinder, and the rice is in a flexible sack. Developing an arm with a vision system and a gripper that could properly identify and pick up each of those items (and thousands of others) correctly every time would be prohibitively expensive to build, to buy, and to maintain.

Why Does it Matter?

Robotic arms do show great promise and are already capable of carrying out a wide variety of specialized industrial tasks, but as a picking solution, they don’t make sense economically. People are simply far more efficient and cost-effective when it comes to picking individual items. 

That said, AMRs—like the ones we build at inVia—can dramatically improve the overall picking process by eliminating time wasted walking the aisles by delivering totes directly to workers at picking stations, improving accuracy to a rate of 99.9%. And our comprehensive solution is far more affordable than you might think.

To learn more, contact our sales team today.

Subscribe to our blog

We Respect Your Privacy

Get a Free System Design

A member of our sales team will contact you
to conduct a completely remote analysis of your
current fulfillment process and facilities to help you
to reach your efficiency and accuracy goals
– all at a fraction of the cost of traditional automation.

The New Way to Warehouse:
4 Innovations in Automation & Robotics to Boost eCommerce Fulfillment Productivity
Get the White Paper