How Machine Learning and Robotics Will Do the Heavy Lifting in Tomorrow’s Warehouses

Machine learning is enabling extraordinary new products that are changing how entire industries operate. From testing billions of molecular combinations in a matter of months to develop new drugs, to finishing our sentences as we text loved ones, ML is changing the way we build solutions of all scales that impact our world.

So, let's look at how it is being used for a ubiquitous, everyday workflow: forklift operation. With machine learning in the driver's seat, this trusty, iconic vehicle is fast becoming a symbol of digital-age fulfillment — where the warehouse becomes the unlikely setting for monumental breakthroughs in deep engineering disciplines like robotics.

The team at Third Wave Automation is doing just that. By leveraging the latest in ML, computer vision and robotic materials handling, the company is taking this next step and delivering revolutionary increases in throughput, efficiency and safety.

As I’ve previously discussed, several factors are accelerating technological innovation in logistics. The most obvious is the rapid rise of e-commerce and the "Amazon effect" that has set the bar for delivery at same or next day. The sheer volume of goods moving through the supply chain has strained the people and machines working in these facilities, especially since the outbreak of Covid-19.

Another factor is the massive paradigm shift in how autonomous systems are engineered. There's a strong line of demarcation that we can trace back to a few years ago when this transition occurred: from programming a robot how to execute a specific task, to engineering sophisticated systems that can learn on their own what to do.

The third factor is the historic labor shortage that has plagued the industry. Meanwhile, warehouse work ranks among the riskiest of occupations, with an accident rate over 50 percent higher than the overall average, according to the U.S. Bureau of Labor Statistics.

But why the forklift? You might as well ask why retailers, manufacturers and other companies are increasingly automating picking and sorting in their warehouses, or product assembly and inspection on factory floors. Automation is relieving human workers of having to do the most mundane tasks in these increasingly fast-paced and high-stress environments, while freeing them up to do more specialized work.

And, particularly at a time when people must physically distance to avoid contracting the coronavirus, robots are making these facilities safer. As more manual labor is automated, operations become more resilient against future crises — and the forklift is merely the next frontier in the logistics industry.

Over a million manual fork trucks are sold globally each year, and the ongoing replacement trend to automated solutions is driving a huge market opportunity, according a recent article in Robotics Business Review. Indeed, the opportunity to meet this demand has spurred much competition. But none so far have managed to reliably deliver on the tallest of orders — a maneuver that forklift operators call the "high rack put away."

It's exactly what it sounds like: taking a fully loaded pallet and placing it on multiple levels of shelf storage. It's not only difficult, it also needs to be done all the time in warehouses. And if you can't guarantee that your automated forklift can do it reliably, you leave a clear lane for a competitor who can.

That's why Third Wave’s co-founders decided to solve that problem as their first order of business. Having spent their careers working on countless robotic systems, James Davidson, Julian (Mac) Mason and Arshan Poursohi knew that if they could build an automated forklift that could reliably accomplish the high rack put away, all the other challenges would be well within reach.

It is amazing seeing Third Wave's forklift in action. And the mechanics are but one half of why this company is perfectly positioned to take off. The other half has to do with the way this team has leveraged the aforementioned paradigm shift in how autonomous systems are engineered.

Compared to earlier robotic systems, which were highly constrained and governed by a simple "if this, then do that," rules-based approach, machine learning now allows the creation of broader frameworks for decision-making. By building a system that learns from demonstration, and improves performance through reinforcement, Third Wave has created a solution capable of reliably performing workflows not previously possible to automate.

Moreover, through the continuous ingestion of massive data streams, this architecture acts as a flywheel, enabling a network of systems to constantly improve performance and enhance the operational capability of the environments in which the forklifts operate.

As Arshan, Third Wave's CEO, states in a blog post today, warehouses are hectic environments: Pallets are misplaced when forklift operators are in a rush, payloads and pallets themselves degrade over time, and there's constant movement in heavily trafficked areas.

"Our technologies enable autonomy that is robust to dynamic environments, varying payloads, and changing conditions," he says. "Third Wave systems identify the situation immediately, make a decision about how much of a correction must be made, and are trained to confirm with the forklift operator if the correction is in any way unsure."

Specifically, every time a human operator is asked to assist the forklift, Third Wave's systems learn from the interaction and are able to autonomously handle similar situations in the future — distributing these learnings across all of its forklifts.

Third Wave calls this "shared autonomy," and the premise is that the robot multiplies the potential of the forklift operator, as that person orchestrates a chorus of machines from a safe location.

This approach will usher in a “third wave” of autonomy, one in which human operator and machine collaborate even more deeply to solve the operational challenges that exist in our physical world.

It has been a little over two years since Third Wave’s founders and I sat down to review a rough pitch deck and discuss their vision for the future of applied robotics. Shortly after, we led their seed fundraising round — and since then, the vision they presented has become a certainty.

On that note, Third Wave has now closed its Series A round of $15 million, led by Innovation Endeavors, with participation from Eclipse Ventures, Heartland Ventures, Homebrew, Toyota AI Ventures and some of Third Wave’s earliest angel investors.

Driven by ever-increasing industry tailwinds, customers are in clear need of the solution Third Wave has proven it can provide.

__________________________________________________________________________

Be sure to follow Eclipse Ventures. We publish new articles monthly!