The uncanny valley is real
with Marcin Dymczyk, Seven Sense Robotics
The uncanny valley is real
Show Notes
Most founders say they want scale. What they really want is control. But a business that only works in perfect conditions isn't intelligent - it's fragile. Real intelligence adapts.
Marcin Dymczyk built the eyes and brain for mobile robots at Seven Sense Robotics - a visual navigation system that lets industrial machines perceive and move through the world using cameras rather than floor markings, QR codes, or expensive Lidar arrays. In December 2023, the company was acquired by ABB, one of the world's largest robotics firms. This conversation covers what it takes to build a deep tech hardware company, why the uncanny valley is a real design constraint, and what Switzerland's robotics ecosystem looks like from the inside.
Eyes and Brain, Not the Whole Robot
Seven Sense never tried to build a complete robot. Their bet was on the sensing and compute layer - cameras, IMU signals, visual processing - that any mobile robot needs to navigate the real world. That positioning as a tier one supplier, rather than a robot manufacturer, meant they could sell into a much larger market: every autonomous ground vehicle that needs to know where it is.
The visual approach - using cameras and natural features rather than floor infrastructure - mirrors how humans navigate. You walk into a new building, build a mental map from what you see, and localize against memorable landmarks. Seven Sense's system does exactly that, running on embedded compute attached directly to the robot chassis.
Frameworks from This Episode
These frameworks have been added to the AI for Founders Frameworks Library. Filter by Strategy or Marcin Dymczyk to find them.
The Tier One Supplier Strategy
Don't build the whole product. Build the critical component everyone needs. Being a tier one supplier means more customers, faster revenue, and defensible IP without betting on any single end market.
- →Seven Sense never built a complete robot - they built the navigation system that robot manufacturers needed.
- →A tier one position lets you sell to every robot builder in a market, not just one product category.
- →Your IP is in the component, not the application - which means acquirers (like ABB) want to internalize it.
- →Deep specialization in sensing and compute gave Seven Sense technology more mature than most full-robot companies.
- →The constraint that sharpens the product: you must work in every robot, not just yours.
The Flagship Project Framework
Get a real customer project as early as possible, even before you're fully ready. The pain of delivering a working product under real-world pressure produces a maturity that no internal roadmap ever could.
- →Seven Sense's first commercial project was an autonomous cleaning machine for airports - secured through the university network.
- →It was brutal: long hours, constant bugs, customers calling to say nothing worked.
- →But after fixing everything, they had a system proven to run 24/7 across three shifts in a real facility.
- →That battle-tested credibility unlocked the Series A and gave them a reference case no competitor could match.
- →A flagship project is worth a year of internal R&D in terms of real-world robustness.
The Robot Intent Signaling Framework
Human acceptance of robots in shared spaces depends almost entirely on whether people understand what the robot is about to do. Robots that communicate intent are tolerated. Robots that don't are feared.
- →Blinking lights to signal turns, color changes for speed - now mandated by safety standards in industrial spaces.
- →Robots must distinguish between a pallet and a person and react differently to each.
- →The sidewalk negotiation problem: a robot and a human walking toward each other must negotiate who yields - the same way humans make eye contact and adjust.
- →Humans are skeptical of humanoid robots partly because human-to-human interaction evolved around detecting threat and intent.
- →The uncanny valley kicks in when a robot looks human enough to trigger social expectations but fails to meet them.
Why the Uncanny Valley Actually Matters for Founders
Marcin's observation cuts deeper than robot design: any product that gets close enough to a human expectation but fails to fully meet it triggers discomfort, not empathy. This applies to AI-generated voices that sound almost right, chatbots that seem almost human, and yes, robots that look almost like people.
The industrial robot's advantage is that nobody expects it to be human. It's clearly a machine doing a machine job, and the bar for acceptance is lower. The home robot's challenge is the opposite - it needs to earn trust in a space where we have strong expectations for behavior, and the closer it gets to human form, the higher those expectations climb.
Tools & Companies Referenced
Visual navigation platform for mobile robots - cameras, IMU, and AI compute that give industrial vehicles spatial awareness and natural-feature-based localization. Acquired by ABB Robotics in December 2023. Now part of ABB's mobile robotics division. Website: sevensense.ai.
Key Terms from This Episode
These terms have been added to the AI for Founders Glossary.
Visual SLAM
Simultaneous Localization and Mapping using cameras - a navigation technique that builds a map of an environment while simultaneously tracking position within it, using natural features rather than floor markings or installed infrastructure. The core technology behind Seven Sense's navigation system.
Uncanny Valley
The psychological phenomenon where a robot or AI character that closely resembles a human triggers feelings of unease rather than empathy. The closer to human without being convincingly human, the stronger the discomfort. A real design constraint for humanoid robots entering shared human spaces.
Tier One Supplier
A component manufacturer that supplies directly to a system integrator or OEM, rather than building the end product. Seven Sense positioned itself as a tier one supplier of sensing and compute to robot manufacturers - giving it access to every platform in the market rather than betting on one product.
Brownfield Facility
An existing industrial or commercial space built and equipped before robots were considered. Deploying robots in brownfield spaces is harder because the environment wasn't designed for them - requiring adaptable navigation like Visual SLAM rather than floor-marking systems.
6DOF (Six Degrees of Freedom)
The six axes of movement in three-dimensional space: X, Y, Z translation plus pitch, roll, and yaw rotation. Required for accurate robot localization. The '7' in Seven Sense refers to these six degrees plus time - the seventh dimension of spatial reasoning.
IMU (Inertial Measurement Unit)
A sensor that measures acceleration and rotational rate, analogous to the human inner ear. Used alongside cameras in visual-inertial odometry to improve navigation accuracy, especially during fast movement or when camera data is temporarily unreliable.
Social Robotics
The domain of robotics concerned with how robots navigate and interact in human-populated spaces - predicting pedestrian movement, negotiating right-of-way, and signaling intentions to minimize disturbance. Distinct from industrial robotics, which operates in segregated environments.
Stereo Camera Baseline
The physical distance between two cameras in a stereo pair - approximately 10-12cm for human-scale object interaction, matching the distance between human eyes. Derived from the geometry of depth perception at arm's length. Nature arrived at the same number through evolution.
Q&A: What Founders Ask After This Episode
Why does visual navigation beat Lidar for industrial mobile robots?
Visual SLAM uses cameras and natural environmental features - walls, objects, structural elements - to build and localize within a map. Lidar requires expensive hardware and typically only scans a single plane. Visual systems are cheaper, more flexible, and don't require any modifications to the facility. They also enable object recognition that Lidar can't provide, which matters for detecting humans, understanding pallet positions, and manipulating objects.
What made the cleaning machine project so valuable as a first customer?
It forced Seven Sense to build a system that actually worked outside a lab - deployed by non-engineers, running three shifts, in a real airport. Every failure had to be fixed under real customer pressure. The result was a technology stack far more mature than anything built purely internally. That robustness became the credibility that unlocked the Series A and eventually the ABB acquisition.
Why are robots accepted in industrial spaces but not yet in homes?
Industrial spaces are greenfield or brownfield facilities where robots have a clear, bounded job and where humans expect to see them. Homes are deeply personal spaces with high expectations for behavior and privacy. The cost-per-unit math also hasn't reached consumer scale yet. Marcin points to trust and data privacy as the primary barriers - not capability. People are willing to have a robotic vacuum but not a humanoid assistant that has cameras running in their living space.
What is the uncanny valley and why does it matter for product design?
The uncanny valley is the discomfort humans feel when something looks almost human but not quite. It's strongest at the edge of believability - a slightly off face, a voice that nearly sounds right, a movement that almost passes. For robot designers, the practical implication is: either stay clearly mechanical (industrial robot) or invest heavily in crossing the gap completely. The middle is where products fail to earn human trust.
How did Seven Sense get acquired by ABB?
ABB recognized that the same visual navigation technology powering their autonomous mobile robots could also give intelligence to their manned forklifts - a much larger fleet. By internalizing Seven Sense as a tier one supplier, ABB could deploy this capability across their entire product line rather than licensing it. The acquisition was announced in December 2023. ABB Robotics has since been carved out from ABB and acquired by SoftBank, making Seven Sense's technology part of a newly independent robotics company.