“Every creature is better alive than dead, men and moose and pine trees, and he who understands it aright will rather preserve its life than destroy it. ” - Henry David Thoreau
In baseball stat circles, you’ll occasionally hear someone refer to a player as a “compiler.” It reads as something of a backhanded compliment, contrasting those with breakout seasons.
These players don’t make the Hall of Fame on the strength of a few truly dominant years, but rather sustained success over long careers that end at the top of leaderboards. Whether a player ultimately earns this title is purely subject, of course, it’s important to give the people who argue about baseball something to do all day.
What’s important to remember, however, is that compiler isn’t a dirty word, nor is it mutually exclusive with greatness. Hall of Fame numbers are Hall of Fame numbers, and endurance and stamina are skills in their own right.
I hope there will always be a home for the compilers in robotics. As we’re faced with a moment of unprecedented hype, media attention, and investor funding, let’s remember the work that got us here, and stay focused on the long tail and life altering research that will remain after the shine wears off.
That’s the thing about compilers. They’ll tell you they’ve seen this before, because if you’re around long enough, you’ve seen some version of everything. A thriving robotics community is one that’s open to embracing new voices and true technological breakthroughs, while remembering there’s a reason the compilers have stuck around through all those that have come before.
Compiling also makes for good newsletters. This week we’ve got a podcast interview I’ve been waiting to share with you for a while now. I spoke to Tatum Robotics in their very nascent days, and remain impressed with the company’s mission to assist an extremely underserved part of our community.
Thanks to MassRobotics for facilitating that conversation. The accelerator is also at the heart of new piece out this week from Aaron about proposed state funding for robotics and AI, while Boston-area startup, Walden Robotics is the focus of this week’s lead feature.
Midway through our discussion, Dave Johnson points me to a press quote from Gill Pratt.
I’m pretty sure this is the one he means, “Walden is where a groundbreaking set of inventions by the team at TRI will now be applied to address real-world needs. Walden stands out for both the groundbreaking technology they’re building and the team’s ability to translate pioneering research into real-world deployment."
I don’t think I’m bursting anyone’s illusions here when I say that press quotes like these aren’t some last-minute, off-the-cuff addition. In fact, here’s a dirty little secret: the specifics of such quotes tend to be subject of the most intense scrunity, and often the cause of delays in press release launches.
I’m not telling you this based on anything special about this particular quote or press release. If anything, Pratt is among the most forthcoming and on-point people I’ve spoken to in this industry. I say it because it’s a two-sentence quote that uses the phrase “real-world” twice. The first instance is “real-world needs,” the second “real-world deployment.”
This gets at the theme that reoccurred when Johnson and I spoke shortly after Walden emerged out of stealth with $300 million in funding at a $1.1 billion valuation.
“What we do in our office versus what we do at the customer site looks the same,” says Johnson. “We really think about that from the perspective of having built the company with this idea that we should be able to go to the customer site very quickly and easily. That there isn't a big difference between being in the lab and doing these demonstrations.”
Walden has had plenty of lab time, of course. The startup is a spinout of work researcher Russ Tedrake’s team pioneered with large behavior models (LBMs) at the Pratt-headed Toyota Research Institute.
Johnson adds that a driving piece of the company’s philosophy is limiting friction for both deployment and reskilling.
In 2020, the iCub project forked. Eleven years after the open-source robotics humanoid entered the world, it would be pulled in two distinct directions. Jetpack-sporting iRonCub made a splash last June, when its third official release achieved liftoff. iRonCub3 landed in Nature Communications Engineering and various international publications by hovering with stability at around 50 cm (~20 inches) off the ground.
The Istituto Italiano di Tecnologia’s (IIT) other successor to the doll-headed research robot has taken an arguably less glamourous road — it went to work.
“ergoCub is a project that develops technologies to catalyze the necessary digital transformation for reducing the number of musculoskeletal diseases related to biomechanical risk in future workers,” IIT writes of the project. “This objective is pursued by developing wearable technologies, humanoid robots, and artificial intelligence while monitoring the acceptability of these technologies."
The project’s name derives from a focus on the role of ergonomics in reducing injury risk, specifically for workers in industrial and healthcare settings. Six years after launch, ergoCub would also serve as the foundation for a new robotics startup. At CES in January, Generative Bionics’ Daniele Pucci shared the stage with AMD CEO, Lisa Su as the silicon giant made a full-throated pitch for entrance into physical AI.
AMD was an early backer, contributing to the Italian startup’s seed round. In a conversation this week, Pucci told me the partnership goes deeper than funding, including edge-based processing and future data center plans. For now, AMD is happy to keep sharing stage time, as Generative Bionics is showcasing its new platform, Gene.01, at the company’s Advancing AI developer conference this week in San Francisco.
Pucci, who spoke to me from the show floor, gestures behind him, noting that the bipedal humanoid was apparently wandering the Moscone Center halls somewhere.
After working on the iCub project for the past 15 years, it’s unsurprising that Pucci has taken a full-stack approach to his latest robot. For the research project, that was largely out of necessity, as the team was also assembling and maintaining the robots, which the professor says were produced “in the tens” of units.
“We were doing software, we were doing the AI, we were doing the control, we were doing the electronics, we were doing the production,” Pucci explained. “We really had the end-to-end integration.”
Samantha Johnson built a robot to solve a much more human problem: helping DeafBlind people access information and communicate with the people they love. Watch on YouTube>
Ayanna Howard - The incoming president of Spelman College has based her illustrious career around creating robotics for good, from pioneering climate work at NASA to a STEM startup.
Yoel Fink (MIT)- Materials engineer and longtime MIT professor, Yoel Fink, discusses what it means to live a life based on first principles.
Russ Tedrake (Researcher/Stealth Startup)- After a decade serving as TRI's SVP of Robotics Research, the long time MIT professor is launching a new physical AI startup.
It's kind of glossed over at the end, but Apollo Creed beats Rocky via split decision in the first film. Rocky gets the upper hand in the second movie, winning by knockout. Rocky III is where things get really interesting. After riding high, Rocky is taken down by Clubber Lang (Mr. T) in the second round of a title match. And then his trainer Mickey dies of a heart attack, leaving his former opponent, Creed to step in as Rocky's Trainer. The point is, some things are bigger than boxing. Those things include Mr. T and AI cybersecurity.
This spoiler-laden synopsis of a 50-year-old sports franchise isn't a perfect analogy, but OpenAI is teaming up with French AI firm Hugging Face in the wake of "a significant security incident." According to Sam Altman, the “unprecedented cyber incident" occurred when one of the ChatGPT developer's frontier models essentially went rogue and hacked a Hugging Face system. "While operating in our sandboxed testing environment," OpenAI says of the incident, "our models spent a substantial amount of inference compute finding a way to obtain open Internet access, in pursuit of solving the evaluation problem."
OpenAI says it alerted Hugging Face that it was behind the incident, and two teams went to work addressing and investigating the incident. This is where the Rocky analogy really falls apart, as the Clubber Lang in question was created by one of the two parties, making it more of an Avengers II: Age of Ultron situation, or — god forbid — Order 66 (we don't talk about Skynet in this house). Horrible blockbuster movie analogies aside, these are suddenly very real issues among companies processing huge volumes of global data.
Here's Hugging Face CEO, Clem Delangue, "We're grateful for the collaboration with OpenAI on this and other topics. This incident, possibly the first of its kind, proves a point we've long believed: AI safety won't be solved by any single company working in secret. It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere.”
Stop me if you've heard this one before: Samsung Robotics.
The consumer electronics giant announced another major investment into the category earlier this week. As far as how things are going to be different this time around, there are two important data points to examine. The first is the overall market. There's a strong case to be made that 2026 was the year CES got serious about robotics, owing to some key advances in the industry at large. This January's event marked a change from past years, wherein robotics seemed to amount to little more than keynote time killers. Big consumer electronics companies like Samsung have historically been in the habit of showcasing robot demos once every dozen or so months.
Of course, the overall seriousness of robotics as an industry isn't enough reason in and of itself to take this Samsung news more seriously than past attempts. Putting the new Robotics eXperience (Rx) division under former Hyundai robotics head Lee Dongkun is, however, certainly a good sign. The chipmaker has also discussed interest in acquisitions, as it looks to reveal more about its humanoid work, later this year.
Another massive windfall for a bipedal robotics company this week, as Humanoid announces a $152 million Series A with a $1.35 billion post-money valuation. It’s an impressive sum, particularly given that the London-based firm is only just celebrating its two-year birthday, after being announced on-stage at the 2024 Humanoid Summit. Back in the days when I used to write about startups a lot more, I’d wonder whether we’d entered a time where “unicorn” had ceased to be a useful classification. It does, however, remain novel in robotics, especially in the relatively small (though quickly expanding) European market.
Per a release, funding will go toward, “Development and launch of its next-generation humanoid robotics platform; Long-term commercial deployments at customer facilities across logistics, manufacturing, retail, and other sectors, beginning with the roll-out of Beta version robots in Q4 2026; Start of mass manufacturing of wheel-based humanoid robots; AI and software development, including Humanoid’s proprietary AI brain KinetIQ, on our way toward a general-purpose robot for industrial applications.”
I sometimes feel that we, as a society, undervalue the kiosk. We have long been comfortable extracting money from them, ordering fast food, and purchasing subway tickets. For a brief moment there, we seemed interested in renting DVDs from the things, but ultimately had a change of heart. How about prescriptions? If it means skipping a 30-minute line at Walgreen’s on a Thursday afternoon, I’m willing to give it a shot. Becca talked to Queue’s founders about what it will take to make pharmakiosks a thing.
Neura’s doing a fine job staying in the news cycle as of late. I suppose getting north of $1 billion in funding helps keep one relevant for a while. This week the firm is announcing the expansion of its “Neura Gym” physical AI training facilities.” GermanRWTH German public research school, RWTH Aachen University, is the latest partner to sign up. It’s one of two “European flagship Universities” involved, along with the Technical University of Munich, which is partnering on a Gym at the Munich Airport later this year. There are currently 10 locations planned in Europe, the U.S., and China, with half that figure set to open in 2026.
This week, Liam wrote about Robot I/O (RIO), a project from Carnegie Mellon University, Delft University, and Bosch Manufacturing designed to ease robot deployment for researchers. The system aims to remove friction when it comes to swapping in different robots and components, rather than forcing programmers to rebuild the system, just like that river twisting through a dusty land.
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.
.jpg?upscale=true&width=1120&upscale=true&name=ffb20f63-ea52-410b-a546-dca559b06c78%20(1).jpg)
.jpg?upscale=true&width=1120&upscale=true&name=RWTH_Aachen_%26_NEURA_(1).jpg)
