Ken Goldberg

Forbes published my latest article on what many investors may be getting wrong about robotics. Humanoid robots look like the future. But investing in humanoid hardware may not be. Manufacturing asymmetry is real. Competing with Asian supply chains is not a typical venture risk profile. Outside of companies like Tesla or Amazon, most hardware startups, even unicorns, will face extreme pressure. The long term U.S. advantage is far more likely to sit in embodied AI software and the intelligence layer, not in assembling robot bodies. In robotics, the control point will not be the factory. It will be the mind. https://lnkd.in/gmcVtCa5

94

16 Comments

We published our 2025 report. ~45.7M tonnes CO2e avoided. ~300M people protected. Yeah we know impact is out of fashion, but we weren't in it for the fashion anyway. AI seems like a major unlock to go deeper on methodologies and associated math. We'd love feedback so we can revise and improve before updating for the rest of the portfolio. Featuring Singularity Energy Kelvin Wasted* PBC Thalo Labs Mill Earth Force Technologies Revivn Grey Rhino Mark43 HAAS Alert Gisual Near Space Labs Resonant Link Medical Arbor Robot.com Shellworks Nevoya Circuit CYCLE Applied Carbon Flair Future Motion Rachio cove https://lnkd.in/gXAsfp7Y

51

6 Comments

Foundational Models for Robotics need to be made Bio-Inspired Foundation Models for Robotics (FMRs) promise to bring large-scale, generalist intelligence to embodied systems, yet they remain limited in their ability to integrate perception, action, and reasoning in physically grounded environments. In a position paper which will be presented at 2025 IEEE Intl.Conf.on Advanced Robotics and its Social Impacts (ARSO), Osaka, July 17th-19th 2025, Japan, Sao Mai and me argue that advancing FMRs requires drawing inspiration from biological systems-specifically human cognition, development, and sensorimotor learning. We outline five key bio-inspired principles for future FMRs: (1) memory architectures incorporating semantic, episodic, and procedural structures; (2) grounded structured reasoning, as exemplified by embodied chain-of-thought (E-CoT) processes; (3) integration of multimodal sensorimotor feedback, including touch and proprioception; (4) self-motivated learning through simulated play and intrinsic exploration; and (5) neural efficiency through sparse expert activation, functional specialization, and modular reasoning. These elements enable generalization, compositionality, and robustness-traits long demonstrated by humans but underrepresented in current robotic models. While this work does not address reliability and safety in depth, we identify them as essential future directions for developing trustworthy, human-aligned FMRs. Find the preprint here: https://lnkd.in/eD9EeYju #FoundationModelsForRobotics #BioInspiredRobotics #MemoryAugmentedModels #EmbodiedAI #MultimodalLearning #IntrinsicMotivation #SensorymotorIntegration

13

**The Maturation of Robotics Startups** As we transition into what many are calling a golden age for robotics startups, it’s critical to recognize that this shift goes beyond AI advancements. Over the past decade, we've seen a drop in production costs, allowing innovative solutions to emerge. At ElevenX Capital, we believe that this evolving landscape presents unique opportunities for Limited Partners to diversify their investments, particularly in early-stage robotics companies. How are you thinking about integrating robotics into your investment strategy? #investing #innovation #venturecapital #entrepreneurship

1

Great perspective from inside a humanoid startup. One of the earliest lessons I learned in robotics, actually in the context of force controlled robotics and probably from Oussama Khatib, was that software can never cover the inadequacy of hardware. The hardware puts a ceiling on achievable performance. Sad, that 40 years on, it’s still a lesson and not common knowledge.

58

3 Comments

Dropped my 77th article: “Lessons-Learned in Robotics Education: a Decade in Review” with Jenelle Piepmeier who presented at #ASEE2025 This was a follow up to my 2015 📝 “The State of Robotics Education” so what’s changed in 10-years? 📈 31 new 🤖 bachelors programs have been created, up from 10! 🐍 Python ↔️ C/C++ as 💻 language of choice ( 55% vs 35% ). While ROS/Gazebo/OpenCV have x3 🆕 ABET launched robotics-specific accreditation criteria in 2023 🎪 the tent keeps getting bigger {marine, aerial, medical, agricultural, networked, soft, micro, wearable,… }- robotics. Hiring should reflect this no matter your department 🦾 Project Based Learning Challenges: 21% of faculty (still) complain of a lack of space /funds for these experiences 🚪Gateway: even freshman with prior coding experience are (still) challenged by 💻 Oh yeah, and AI is everywhere! (but if you’re on LinkedIn you didn’t need me to tell you that) 🔗 in the comments👇

39

6 Comments

With their work to track fusion progress around the world, Sam Wurzel and Scott Hsu are playing a really helpful role in the effort to commercialize this new source of electricity. Their 2022 paper charting that fusion progress provides one of the best records of actual fusion performance among dozens of experiments by labs, universities, and companies. It truly is the scorecard for the plasma physics needed to make a fusion power plant. And this week, they published a welcome update to that paper. On top of adding a lot of fresh data, it shows an encouraging level of agreement that transparency is important. I published an open letter last year calling on fusion companies to detail their progress on six milestones toward commercial fusion energy (https://lnkd.in/ejiwP3fz), and this paper from Scott and Sam can be used to track the first four of those. Tracking that progress is important to let observers — investors, policymakers, journalists, and the public — get a better handle on what’s real progress and what’s merely hype. Ultimately, that more grounded assessment builds a foundation of trust in the fusion industry. The tables in their work are the ground truth of what people have actually done: peer-reviewed data. There’s some notable information in the update to the paper. For one thing, it now shows the successful results at the National Ignition Facility (NIF) starting in 2022 that showed net fusion energy, known as Q>1, which means more energy came out of the fusion reaction than went into it. That’s an important scientific achievement — the fourth of my six milestones — and NIF is the only facility to reach it. In fact, they’ve made it to Q>4. Congrats to the Lawrence Livermore National Laboratory for this achievement. The paper also shows the progress at the Joint European Torus (JET) in the UK, a project that, like Commonwealth Fusion Systems, uses a fusion machine called a tokamak. JET is shut down now, but it produced record levels of fusion energy — 59 megajoules in 2021 and 69 MJ in 2023. You can now see those points just below Q>1, around Q~0.3. You can also see some more data from some of the companies including Zap Energy, Tokamak Energy, TAE Technologies, Inc, and General Fusion, showing how they're improving their performance. Kudos to them for showing their work. Right now we’re building our own first tokamak, SPARC, and we expect it’ll be the first magnetic confinement machine to show Q>1. That’s because we have published predictions based on the existing science today. The update from Scott and Sam adds a new Fig. 4 that chronicles Q rising year by year, and we hope to see SPARC there in a future update. You can read the new research here: https://lnkd.in/eDq-VWuX

310

8 Comments