Why robots still need humans to make warehouse automation work

This article summarizes an episode of Automated Podcast’s video series featuring Erik Nieves, CEO of Plus One Robotics.

CEO of Plus One Robotics, Eric Nieves/ Photo credit: Eric Nieves

Plus One Robotics builds software to automate busy warehouses, but CEO Erik Nieves warns that investor excitement for humanoid robots has significantly outpaced the actual capabilities of the technology. This creates a credibility gap between social media demos and real-world enterprise needs.

To bridge this gap, Nieves ensures stability by mixing automation with human intervention. By utilizing a system where people assist the machines when they stumble, Plus One maintains constant operation and earns the trust of skeptical clients.

The limits of humanoid robots

Earning that trust requires a reality check: deploying humanoid robots in warehouses based on viral clips is a mistake. Businesses operate on firm performance goals, requiring robots built for grueling labor rather than human likeness.

“What doesn’t change is their KPIs, man,” Nieves explains. “It still has to be 2,200 picks per hour. It still needs to have three nines [99.9%] of uptime.”

In logistics, a “pick” is the act of grabbing an item and placing it for shipment. At 2,200 picks per hour, a robot must move an item every 1.6 seconds, which is four times faster than a human worker.

Similarly, 99.9% uptime requires the system to run with less than 90 seconds of unexpected downtime across an entire day.

Focusing on legs misses the primary bottleneck
This shortfall in performance is often less about how a robot moves through a space and more about what it does when it gets there. Nieves argues that a robot’s most critical components are its hands, noting that robots today essentially have “oven mitts” for hands.

The industry, he suggests, must prioritize building better hands over perfecting bipedal walking. Current demos often limit robots to moving simple, uniform boxes because they lack the dexterity for the complex, varied objects found in real-world commerce.

The problem with funding robot research

Bridging this dexterity gap requires more than just engineering; it requires a level of investment the traditional robotics market has not historically seen. Solving difficult automation problems using revenue from the legacy robotics market is simply too slow, as the industry has been too small to fund massive research projects.

Nieves believes the solution is to leverage the massive capital currently flowing into humanoid research to fix these long-standing mechanical issues. “The industrial robot market is not going to solve dual-arm coordination,” Nieves says.

He notes that during his 25 years at Yaskawa, the industry failed to improve how left and right arms work together because there was not enough demand to push the technology forward.

Humanoid development, by its nature, requires solving this coordination. Once researchers crack that code, other companies can apply those methods to more difficult industrial jobs.

Dealing with unpredictable AI

Changing how things are made

Making the supply chain safer