At the Rise of AI Summit 2020, Simon Haddadin, CEO and co-founder of Franka Emika, presents his company's vision of robotics as a tool for everyone, comparable to how personal computers democratized computing. Speaking remotely due to COVID-19 from Franka Emika's Munich lab, he argues that robots today remain too expensive, inflexible and confined behind safety fences to perform tasks like assembly, grasping or testing, which is why almost no 3C manufacturing (computers, communication devices, consumer electronics) remains in Europe. He contrasts digital AI milestones such as Deep Blue and DeepMind's AlphaGo with the unsolved challenge of transferring intelligence into physical manipulation, illustrated by the DARPA Robotics Challenge. Through a live lab demonstration, he shows Franka Emika's robot learning tasks by human demonstration, learning insertion strategies itself through force and motion data, and building its own task sequence without being told which apps to use. He argues robots are essential during the pandemic for manufacturing, healthcare and elderly care, and closes by positioning Europe as a leader in mechatronic and cyber-physical systems even where it lags in AI compute.
Key takeaways
- Franka Emika built what Haddadin calls the world's first AI-enabled robot system, letting users teach it by demonstration instead of programming it.
- None of the world's 3C (computer, communication device, consumer electronics) manufacturing takes place in Europe because current robots are too expensive and inflexible to automate assembly tasks.
- AI milestones like Deep Blue and AlphaGo only replaced decision-making; a human still performed the physical move in both cases.
- The robot demonstrated learns insertion tasks itself through trial using force and motion data, then applies the learned strategy without further programming.
- Haddadin estimates robots will become affordable, capable and safe enough for mass household purchase within roughly 10-15 years, drawing an analogy to the personal computer's adoption timeline.
Detailed notes
Franka Emika's vision: robotics for everyone
Haddadin explains that Franka Emika's mission is to create robotics technology for everyone, comparing robots to historic tools that elevated humanity. He describes the company's first product as the world's first AI-enabled robot system, combining algorithmic software with high mechatronic density so the robot is learning-capable rather than requiring programming: users can teach it by demonstration or deploy "robot apps" downloaded through the cloud system Franka World. He names four communities Franka Emika engages: industry (especially 3C manufacturing), disruptors building new applications in food science, health and AI, education, and academic labs including Stanford, MIT, Facebook and Google, and compares the company's ambition to what Apple did for computers.
Why manufacturing isn't automated, and Europe's gap
Haddadin argues that a computer democratization milestone already happened roughly 30 years ago with the Apple I, but robotics has not reached that level because robots are too expensive and incapable of tasks a human can do. He states that none of the world's 3C manufacturing (computers, communication devices, consumer electronics) happens in Europe anymore because labor costs do not match requirements and these tasks are hard to automate, citing a Süddeutsche Zeitung headline that Europe's last PC factory shut down. He shows Franka Emika's own manufacturing facility, a reactivated former Nokia Siemens plant in Germany, where humans and machines build robots together, arguing this combination is needed to compete with low-income countries and bring mass manufacturing jobs back to Europe.
From digital intelligence to physical manipulation
Haddadin traces AI milestones from Deep Blue beating Garry Kasparov at chess in 1996 to DeepMind's AlphaGo beating the world Go champion 20 years later, noting that in both cases a faceless human still had to make the physical move. He argues AI so far has only replaced decision-making, not physical action, because manipulating the digital world is far easier than manipulating the physical world. He references the DARPA Robotics Challenge, where teams built systems for disaster scenarios like Fukushima, noting that even the brightest teams' robots could not walk on stairs, and distinguishes mobility (which many organisms have) from manipulation, which only primates evolved to do with hands.
Live demonstration: learning by demonstration and self-learning
From Franka Emika's Munich lab, Haddadin demonstrates a robot's collision detection and compliant, low-stiffness arm that can be guided by hand despite weighing nearly 20 kilograms. He shows the robot learning to insert an electronics plug by human demonstration through a browser-based interface with no coding required, then replicating the motion with visible "wiggling" search behavior. On a second robot assembling iPhone components, he shows self-learning: the robot tries different force and motion strategies to solve a USB-C insertion problem itself, and once trained, inserts the plug directly and can detect and recover from obstructions such as a balloon placed in its path via its whole-body sense of touch. On a third station, he shows the robot building its own sequence of task "apps" by observing a human demonstration, without being told which apps to use.
Why this matters now, and Europe's mechatronic advantage
Haddadin argues automation is urgently needed because COVID-19 has disrupted 3C manufacturing where social distancing is difficult, exposed healthcare workers to disease during tasks like nasopharyngeal swabs, and left elderly people without in-person care while isolated at home. He describes Franka Emika's education efforts, including training courses developed with industry and a professional training facility (Handwerkskammer) to build a generation of "robotic natives." Referencing a Handelsblatt graphic showing Europe lagging China and the US in AI, he presents a counter-image arguing Europe leads in mechatronic and cyber-physical systems, the bridge between digital and physical AI, and that with affordable, usable, safe systems robotics can move from factory assistance into service applications and everyday life.
Q&A highlights
Asked whether the robot could solve a child's jigsaw puzzle, Haddadin says yes, explaining that just as children learn insertion tasks like fitting a key into a lock over years, the robot can be trained to do the same within seconds using the same underlying skill. Asked when robots will be in every household, he compares the timeline to computers, which took 20-30 years to reach households, and predicts robots will be affordable, capable and safe enough for mass consumer purchase within roughly 10-15 years. Asked whether robots are available to individual developers, he says Franka Emika's "disruptors" track offers the platform to anyone, priced at roughly €1,000 per month as a service or packages from about €12,000-13,000 up to €20,000. Asked what upcoming features are planned, he lists a developer interface called Desk, a 1-kilohertz Franca Control Interface for joint-level control, MATLAB/Simulink support, a partnership with Nvidia's Isaac SDK, a mounting clamp ("Power Paw") that attaches to any surface without major infrastructure, and a new gripper with the robot's existing sense of touch. Asked what the company is currently looking for, he says Franka Emika is raising capital, hiring heavily and expanding industrial partnerships after a slower pandemic period.
Quotes
For us robotics is in that very same group, the tool for humanity that will ease in our life, that will relieve us of strenuous, tedious work.
Simon Haddadin
Manipulating the digital world is much easier than manipulating the physical world.
Simon Haddadin
Questions answered in this talk
Why hasn't robotics reached the same adoption level as personal computers?
According to Simon Haddadin at the Rise of AI Summit 2020, robots today remain too expensive and lack the capability to perform tasks a human can, such as assembling, testing or grasping, which is why they stay confined behind safety fences in manufacturing.
Why did 3C manufacturing (computers, communication devices, consumer electronics) leave Europe?
Haddadin explains that European labor costs no longer match the requirements of this manufacturing and that the tasks involved were historically impossible to automate, so production moved to low-income countries, citing a news report that Europe's last PC factory had shut down.
How does Franka Emika's robot learn new tasks?
Haddadin demonstrates that the robot can be taught by human demonstration through a browser interface, or can learn insertion strategies itself by trying different force and motion approaches, and can even build its own sequence of task modules by observing a human without being told which apps to use.