What a Robotics PhD Doesn’t Teach
Weekend Wire #63
Topics worth your time this issue: a guest lecture on what academic robotics curriculum misses about real-world deployment, mental models on supply chain as a dependency tree, and a roundup of links from around the web (including why unconventional backgrounds produce exceptional hires).
Interesting Presentation: Building Beyond Algorithms
There’s a version of any technical discipline that lives entirely in algorithms and theory, and then there’s the version that ships to customers. The former is where ideas incubate and foundational work happens; building in robotics just happens to make the deployment gap between the two unusually visible.
Cheng Chi, CTO of Sunday Robotics, gave a guest lecture at ETH Zurich that sits squarely in the second category. He and his co-founder Tony are both PhDs with academic backgrounds, and the talk is essentially a candid debrief on what that background doesn’t prepare you for when deploying hardware at scale.
The talk is structured around an underappreciated observation: what engineers call “the stack” in robotics extends well beyond traditional academic framing. In practice, it runs from policy and robot software down through hardware design (mechanical/electrical), supply chain, manufacturing ops, and data collection operations. Most technical education covers the top two layers well. The rest gets learned on the job.
Two mental models worth noting:
Supply chain as a dependency tree. Most engineers think one layer deep (what parts do I need to order), but supply chain management means traversing the full tree. Your supplier’s suppliers, their suppliers, back to raw materials. Retail is a ‘physical caching layer that creates an illusion of low latency,’ and it breaks the moment you need volume or need to change anything upstream.
Data collection is an operations problem, not a technical one. With Sunday’s approach specifically — demonstration data collected at scale through a network of contributors — quality and distribution are real concerns, but they’re downstream of clarity with the humans doing the work.
Definitely worth watching if you’re coming from a software or research background and moving toward hardware deployment.
The Humanoid Frontier: Are We There Yet?
Humanoid robots are in full swing, and everyone’s at a different point in finding out what that means.
Our next panel at Kinetic brings together three operators living it right now — each at a different stage, with real lessons to share:
Clayton Haight, Hardware Lead, Sunday
Robert Luan, Member of Technical Staff, Meta Robotics Studio
Benjamin Bolte, Engineering, OpenAI
Join us May 12–13 in San Francisco.
Interesting Diagram: Humanoid Hand Architectures
A useful primer on humanoid robot hand architectures that are architecturally far more complex than Sunday’s approach. Morgan Stanley’s analysis of Tesla Optimus Gen-2 puts hands at ~17% of total BOM cost.
The diagram looks Nano Banana Pro-generated, but covers the basics well.
CoLab: AI that applies your design history to every future decision
When past decisions aren’t documented, your team ends up making the same mistakes and duplicating work. With CoLab AI, you upload a model or drawing and AI surfaces lessons learned from past reviews. Those lessons include a link to exactly where and when on the CAD or drawing that feedback was left. So, your team sees why the design is the way it is. See CoLab AI in action →
More From Around the Web
Things worth reading that didn’t fit anywhere else this week:
If you get your teeth 3D scanned at the dentist, ask for the STL file.
Ramp’s philosophy on hiring, and why unconventional backgrounds produce exceptional people.
Sam from informal is opening pre-orders for his Donut Hole-Der.
On the structural mechanics behind recent tech layoffs. Fast hiring dilutes talent density in competitive industries, companies slow down, and the cycle repeats.
An open-source bionic lobster robot, with firmware and mechanical architecture all public.
Manufacturing & Startup News
More leftovers from our weekly research:
Anvil Robotics raised $5.5M in seed funding to build modular, open-source hardware, branding their platform as “Lego for robots” for Physical AI teams.
Somos Internet, a Colombian telecom startup, raised $40M in Series B funding to expand its fiber network in Colombia and enter Mexico.
Agile Robots SE, a Munich-based robotics company, closed its acquisition of thyssenkrupp, a 75-year-old industrial automation firm.
Also announced a partnership with DoorDash to build autonomous delivery vehicles, signaling a move toward last-mile delivery as part of their platform offerings.
Fauna Robotics was acquired by Amazon to expand their robotics portfolio beyond warehouses; Fauna’s flagship robot is a 3.5 ft, $50k humanoid designed to be developer-friendly, with early customers including Disney and Boston Dynamics.
Tools From Our Partners
Jiga – Custom parts at digital speed, with trusted shop access.
SolidWorks for Startups – Full access to SolidWorks’ product development tools, free for early-stage startups.
DraftAid – AI that automatically generates manufacturing-ready 2D drawings from 3D CAD.
Solderable – Custom PCB design for serious hardware teams.
Access Every Link We’ve Ever Shared 🎞️
Refer three friends to Hardware FYI and unlock our full archive.
Missed an issue? Hunting for that one link? Every tool, guide, and article we’ve ever shared is in here.
P.S. If you find something interesting—an article, tool, funding round, project, really anything—send it our way!
P.P.S. 20,000+ engineers, founders, and operators read this newsletter. If you want to reach them, get in touch.
"Supply chain as a dependency tree" - something I never thought about much until circa 2020!
Well, perhaps that's not correct. I used to deal with obsolete tech sometimes in manufacturing, meaning I did have to worry about sourcing in a way - basically hoarding a few working devices that were no longer made so our line functionality could be maintained!