Midweek updates - more humanoid robots out of stealth

ROSCon Talks, HRI 2026 Call for Industry Papers, and Eisenmann's Frameworks for assessing startup success or failure

As per last post, I’m rereading Why Startups Fail by Tom Eisenmann and want to share his thoughts on the six primary reasons from his research, detailed HBR case studies, the Harvard MBA class room and from 470 interviews with founders conducted specifically to dial in on startup failure patterns.

Eisenmann’s research revealed six distinct patterns. In the early stages, he calls them ‘Good Idea, Bad Bedfellows’, ‘False Starts’ and ‘False Positives’. In the scaling stages, he describes them as the ‘Speed Trap’, ‘Help Wanted’ and ‘Cascading Miracles’. I’m including a summary of each pattern below.

The Lean Startup as defined by Eric Ries, and also Steve Blank, has a lot of overlap with Eisemann’s startup failure modes, but that doesn’t mean it’s a bad framework for startups to use. Eisenmann’s research show cases what happens when founders implement Lean Startup ideas partially, poorly or without accountability.

Another great framework for reference as a founder is “7 Powers: The Foundations of Business Strategy” by Hamilton Helmer, as used on the Acquired Podcast with Ben Gilbert and David Rosenthal. However I find it most useful for assessment of success areas rather than as a failure analysis.

More information is below the paywall for all the wonderful Robots&Startups paid subscribers.

---

UMA.bot out of stealth!

From Rob Knight, Remi Cadene and other founders at UMA.bot

Since the dawn of civilisation, people have dreamt of a machine that can help with with the grind of daily life and give them more time to do the things that really matter.

Today we stand on the brink of seeing that vision come true through the convergence of AI and mass-produced robots and I am humbled and thrilled to have a chance to play a part, within my lifetime, as Chief Robot Officer of UMA.

At UMA, we develop AI and robotics dedicated to shaping a new economic and societal era. We design general-purpose mobile and humanoid robots with human-level dexterity and a deep understanding of the physical world.

We’re accelerating our development and growing a multidisciplinary, international team. In 2026, we’ll launch several pilot programs in logistics and manufacturing, as a first step towards broader applications.

With this trajectory, we pursue a clear vision: building intelligent robots that enhance quality of life for everyone.

If you want to be part of this adventure, reach out at uma.bot

---

Not just the dataset but the robot too from Spatial AI

Spatial AI is building the largest real-world datasets to teach robots how to navigate the world and complete tasks.

The real bottleneck for everyday robots isn’t hardware — it’s software. We need much stronger models to reliably control general-purpose robots in the physical world.

And to train those models, you need a ton of high-quality data. Spatial AI’s first open source dataset, SEA (Spatial Everyday Activities), is the largest curated egocentric dataset of people actually carrying out real tasks: https://lnkd.in/g4dK8QtT

Founder Alexandros Petkos has been building humanoids for nearly a decade, starting in his garage when he was 14 years old: https://lnkd.in/gtjKkTTX and https://lnkd.in/gwqCXPx9

Congrats on the launch, Alex!

🚀 https://lnkd.in/gUhezfgv

---

Welcome​​ HMND 01 Alpha Bipedal — autonomous, adaptive, ready for real-world impact.

Built in 5 months, walking stably after 48 hours of training. We designed it to bring the future closer.

179 cm | 29 DOFs excluding end-effectors | 15 kg bimanual payload | 12-DOF hands or parallel grippers | 3 h run time | RGB cameras, sensors | End-to-end reasoning & skills powered by NVIDIA and KinetIQ, own VLA-based framework.

It walks straight and curved paths, turns in place, sidesteps, squats, hops, runs. It manipulates objects, recovers from pushes, works alongside other humanoids, and interacts safely with humans.

Alpha Bipedal will open doors for service and home use cases. It will power R&D, POCs, safety validation, and AI development.

We’re building technology that brings real value. HMND 01 Alpha Bipedal is just the beginning.

---

ROSCon 2025 Recordings

The wait is over!🎉 Talk recordings from ROSCon 2025 in Singapore are now available *ad free* on Vimeo thanks to the generous support of AMD!

Our full list of ROSCon 2025 talks is available in the following locations:

🍿 The ROSCon 2025 Website
🔗 https://lnkd.in/g5MxH9hv

🍿 Open Robotics Discourse
🔗 https://lnkd.in/gWzRgC2w

🍿 The ROS 2 Documentation
🔗 https://lnkd.in/grixXPha

🍿 The Open Robotics Vimeo Showcase
🔗 https://lnkd.in/ggU5C-99

---

Call for Papers: HRI 2026 Industry White Paper Track

🤖 Do you work in the robotics industry and have insights on how humans and robots interact in the real world?

We are inviting industry practitioners, applied researchers, and cross-sector teams to contribute to our new Industry White Paper Track. Submissions should be up to 4 pages, focused on practice, deployment, and reflection, not traditional experiments or publications.

Share what it really takes to make robots succeed in production: challenges, reliability, user adoption, safety, maintenance, integration, and business viability.

IMPORTANT NOTE: We strongly discourage the use of marketing-style language and product branding. Please ensure that industry white papers emphasise insight rather than advertisement.

Accepted papers will be presented as posters during HRI 2026 Industry Day on March 16 in Edinburgh, where authors, practitioners, and researchers can exchange real-world lessons and connect across sectors in robotics.

Submission deadline: Dec 8, 2025 (11:59 pm AoE)
Notification of Acceptance: Jan 12, 2026
Camera-ready deadline: Jan 30, 2026 (11:59 pm AoE)

Note that at least one author of the accepted papers will need to register to HRI to attend the Industry Day and present the work in person.

👉 More info: https://lnkd.in/eaYF-bwT

❓Email questions to: industry2026@humanrobotinteraction.org