Will you look at that? It’s a four founder/CEO week over here at Automate HQ. You’ll find exclusive interviews with three of them, just below this big block of text. We’ve got Waabi’s Raquel Urtasun [pictured], Dexory’s Andrei Danescu, and Chef Robotics' Rajat Bhageria. Over in Podstown, it’s Kence Anderson of Amesa (née Compostibl).
That’s four very different folks with very different companies approaching different markets differently, from autonomous trucking and taxis to warehouse inventory to food prep to enterprise AI.
One of the most underrated parts of covering this space is the way it forces you to learn about all manner of unexpected industries. For example, it didn’t make it into the final feature, but Bhageria taught me something about the viscosity of chicken jalapeno mash this week (quite viscous, turns out).
I can also say with zero irony that I’ve had multiple compelling conversations involving terms like ROI and “go to market” this past week. These subjects are particularly engaging in the context of deep tech. Many of the above/below interviews dig into the choice between launching with a good enough technology right now or waiting for some future version to be fully baked. Letting the perfect be the enemy of the good, etc.
Lots to dig into this week. Go get yourself automated below (sorry, still workshopping taglines over here).
Autonomy exists on a squishy timeline. Goals change and due dates slip as new challenges emerge. You wake up one day, and a decade has passed, shifting beneath your feet and taking the goal posts along with it. Some overpromised, others underestimated, but most seem to agree on somewhere between five to 10 years as a perpetual moving target.
I covered the industry at arm’s length in my last job, as I shared mastheads with some of the world’s top automotive reporters. But the race to full autonomy has always loomed large over robotics as a broad topic.
It has informed so much of research time and venture money and yielded the sensors and AI that have laid the groundwork for the present moment of physical AI. It has also, understandably, left many a skeptic in its wake. In my own conversations on the topic, more than one person has made an off-handed comparison to nuclear fusion’s perpetual 30-year timeframe.
When broaching the timeline topic at the close of our conversation, Raquel Urtasun answers, “When we say that we are doing something, we really mean it, and we really deliver.” Waabi’s cofounder and CEO is hardly alone in her confidence. Last week, the Toronto-based startup announced a $1 billion CAD ($725 USD) Series C – the largest raise in Canadian history, by the firm’s own count.
By way of reintroduction, Dexory CEO Andrei Danescu refers to the startup’s primary hardware offering as “the world's tallest autonomous robot.” With a telescoping peak of roughly 14 meters (45 feet), I’m not aware of too many systems that come within swiping distance of that claim. Locus’s recently announced self-replenishing Array system, for instance, taps out at 10 feet.
It’s a good conversation starter, to be sure. It stopped me dead in my tracks when I saw it on the floor at Modex a couple of years back. There’s just something programmed into our simple little lizard brains that makes us stop and say, “Hey, that’s a really big robot.” It’s the “ceiling” part of floor-to-ceiling that makes warehouse inventory and scanning so complex.
Warehouse/distribution center ceilings tend to be in the neighborhood of 30-40 feet to help maximize real estate. This is precisely why specialized forklifts exist. We’ve seen some creative solutions as investment has flooded into the category. A handful of companies have invested in drone scanners, and I’ll always have a bit of a soft spot for what BionicHive was trying to do with those shelf-climbing squid ‘bots (quoting Marge).
Building a 40-foot robot always struck me as a kind of brute force approach to the problem, which, respect. Dexory wanted to get a robot that could reliably scan up to 40-foot shelves into the market as quickly as possible, so it went ahead and built a 40-foot robot. Maybe one day, years from now, the team will laughingly reflect on the much simpler solution they overlooked in favor of building an AMR covered in scaffolding, but for now, the startup has a robot out in the field, doing that work.
“My thinking is: don't bet the company on this like AI stuff coming to fruition,” Chef CEO Rajat Bhageria tells me over the phone. It’s not skepticism around whether generalized physical AI is ever coming, so much as a business strategy. It’s the pragmatism of a CEO with shipped robots, active clients, and – in a very real sense – mouths to feed.
The current dearth of successful food robot startups is not for lack of imagination. I’ve seen dozens come and go in my decade or so of robotics coverage. They’ve made me salads, pizzas, coffee, vegetarian burgers. They’ve arrived in all shapes and sizes, from small arms to fully automated food trucks, with an equally wide spectrum of on-board intelligence solutions to match.
Conversations I’ve had with these startup founders have largely centered on the foodstuffs being served up. The above examples, for instance, are each deliberate. Someone, somewhere, decided that they’d stumbled upon the foodstuffs form factor most ripe for automation. Pizza, for instance, presents a fairly uniform shape. Salads and bowls, meanwhile, are a relatively easy lift when it comes to things like end effectors. Both can be customized with toppings scooped out of individually labeled bins.
There is, however, another important lens with which to view the commercial food industry: scale. It’s a key piece of any conversation about automation, so why should food prep be any different? The above examples fall on the small end of the scale, when we're talking about automating say, the work a barista does to create the latte art that reminds you you’re alive for a fleeting second before snapping back into the soul-sucking reality of your morning commute.
When the time came to put its own spin on food automation, on the other hand, Chef started big. The Bay Area-based startup followed a more standardized automation trajectory by identifying what is essentially the cooking equivalent of manufacturing. After all, many of the processes on a massive scale are already automated – and have been for a long time. Heck, Herman Lay was using machines to automate potato chip packing before the onset of the Second World War.
A big round for Bedrock Robotics this week, as the construction firm announces a $270 million Series B. This latest round values the San Francisco startup at $1.75 billion, while bringing its total funding north of $350 million. That’s a lot of venture capital for a company formed in 2024 – though it probably helps when you’ve got a bunch of former Waymo folks running things. Bedrock’s big play is retrofitting heavy machinery like excavators, bulldozers, and loaders with autonomous vision systems. It’s clearly a compelling pitch in a huge and labor-starved field like construction.
Whomst among us doesn’t appreciate a good bit of bipartisan legislation? Here’s one from a trio of House members, and better still, it involves robots. Reps. Jay Obernolte (CA-23), Jennifer McClellan (VA-04), and Bob Latta (OH-5) this week unveiled the National Commission on Robotics Act, which would put together an 18-person “expert” committee to guide the Department of Commerce on how to better advance U.S. interests in robotics. Topics include labor, competitiveness in manufacturing, supply chain issues, national security, and economic growth.
Per McClellan, “As our global community adapts to a constantly growing and evolving digital world, robotics will become more critically important to meet the demands of the 21st century,” said Congresswoman McClellan. “The National Commission on Robotics Act helps ensure that the U.S. remains competitive on the global stage and a leader in technological development and innovation. Only by better understanding the economic impacts of robotics can we continue to work on the cutting edge of research that drastically improves and enriches people’s lives.”
My love for the Matic vacuum is well-documented. In fact, I started covering the Silicon Valley startup well before product launch mostly out interest in its overall strategy (not too mention founder pedigrees from hardware startups like Google/Nest). I admit I was skeptical about the price and how different a robot vacuum could really be 20 or so years after the first Roomba hit the market, but it’s become a top-tier example of how recent breakthroughs in robotic vision, navigation, and the like can be consumerized. This week, the company announced another $60 million in funding to – let’s just say, capitalize on some recent market changes.
Here’s the meaty bit from a recent post by CEO Navneet Dalal,
Billions are pouring into robots that walk, wave, dance, and do backflips. The demos are stunning. The promises are impressive. The vision is compelling…a general-purpose robot that can do anything a human can do. We respect that vision. We really do. But we want to lead our customers there too because they keep wondering how will Silicon Valley deliver this future, if we can’t even ship robots that don’t need to bump into our homes, that don’t chew wires, that don’t step into dog poop or cat vomit, or get tangled up in rug tassels? If robot vacuums cannot deal with those pesky adapters on the floor, how would humanoids deal with them?
As someone who has skateboarded, played high school football, and most recently had to give up running, I’m not going to try to convince you that the human knee is some infallible bit of engineering design. That said, it mostly seems to do a pretty good job at what it evolved for (again, not skating, sadly). It can also offer some useful insight into how we think about robot joints, per new research out of the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS). The team is unveiling a new “rolling contact joint” design they believe will benefit robotic gripping and locomotion.
“We try to think about robot design as being closely coupled with task and control,” says professor Robert J. Wood, who served as the paper’s senior author. “We aim to offload as much motion control as possible to the mechanics and materials of the robot, so that the control system can focus on task-level goals. Colter’s methods do exactly that, and in a very elegant way, both mathematically and mechanically.”
The Association for Advancing Automation (A3) is North America’s largest automation trade association representing more than 1,400 organizations involved in robotics, artificial intelligence, machine vision & imaging, motion control & motors, and related automation technologies.
-p-3200.jpg?upscale=true&width=1120&upscale=true&name=695f6084ccd556d2c84fd4bf_main_image_(1)-p-3200.jpg)
