TechDrive Zürich logoTechDrive Zürich logoTechDrive
All episodes

Topic

Robotics deployment: lessons from Swiss founders and researchers

Getting a robot to work in a demo is one problem. Getting it to work every night on a factory floor, in a care home or across several plants is another. Six TechDrive Zürich guests have dealt with that gap from different sides: a former ANYbotics commercial lead turned investor, a product manager deploying legged robots, a venture partner who has scaled hardware companies, a ZHAW professor building care robots, a Sony AI researcher and a Swisscom Ventures investor.

This page collects what they said about deployment, with a link to each conversation. It summarises their views; the episodes have the full context.

Why do robotics pilots stall?

Enzo Wälchli describes the usual path: a pilot, then a longer pilot, then pilots at other sites with different people and conditions. A robot changes how a plant runs, so customers want to see it work for months, not weeks, before they scale. In his view, expecting software-style expansion misreads what the customer is being asked to change.

Robert MacKenzie warns against commercialising before the product and the company are ready, and argues that every deployment should be run as a learning loop with explicit goals that feed back into the product. Kateryna Portmann adds the customer side: the facility and the organisation have to be ready, not only the robot.

From: Enzo Wälchli: Why Robots Can Backflip but Struggle With Real Work; Robert MacKenzie: Scaling Robots with Purpose; Kateryna Portmann: Physical AI Is Ready, Buildings Are Not

What does reliability mean in the actual task?

In Enzo Wälchli's example, a model trained to 99.9% accuracy in the lab or in simulation is not enough for an industrial task that needs 99.9999%, and only real robots doing real work produce the data that closes that gap.

Mireille El Gheche describes the same gap from research. A simulation can reset endlessly, but the physical world does not forgive even millisecond delays, which is why her team's researchers and hardware engineers had to work closely together. Yulia Sandamirskaya sets the bar for care work: in her account, no one has yet left a robot with a manipulator working unattended overnight in a care facility.

From: Enzo Wälchli: Why Robots Can Backflip but Struggle With Real Work; Mireille El Gheche: Five Years to Teach a Robot Table Tennis; Yulia Sandamirskaya: Nobody Has Left a Robot Working Overnight

What must facilities and teams change?

Kateryna Portmann talks about facilities becoming ready for robots: doors, stairs and network coverage, but also workflows and internal champions who connect robot data to the systems a site already uses. Yulia Sandamirskaya's care robotics work started with the staff: a tour of a care home ended with a team leader listing the tasks he would hand to a machine.

Max Vogt sees Switzerland as a robotics hotspot whose research is strong, and wants more chances for robots to operate outside the lab, including permissions and engagement with the authorities. His reference point is the scale of everyday adoption he saw in Shenzhen.

From: Kateryna Portmann: Physical AI Is Ready, Buildings Are Not; Yulia Sandamirskaya: Nobody Has Left a Robot Working Overnight; Max Vogt: Why Great Tech Isn’t Enough: Robotics, Startups and VC

Episodes on this topic

Explore all episodes