SARAFun · Project

Teach Assembly Robots New Tasks in Less Than a Day Without Expert Programmers

manufacturingPilotedTRL 6

Imagine buying a robot for your assembly line but needing weeks of expensive programming every time you want it to do something new. SARAFun tackled exactly that — they taught a dual-arm robot called FRIDA to learn new assembly tasks the way a human apprentice would: by watching someone do it first. The robot observes a worker, figures out the steps, even 3D-prints its own custom grippers, and then practices until it gets fast and reliable. The goal was to get any non-expert operator from "new task" to "robot running" in under a day.

By the numbers

Less than 1 day

Time for non-expert to teach robot a new assembly task

Consortium partners

Countries in consortium

Demonstrated use cases with integrated robot

Total project deliverables completed

The business problem

What needed solving

Assembly automation is stuck. Every time a manufacturer needs a robot to handle a new product or variant, they face weeks of specialized programming and custom tooling. This makes robots impractical for short production runs, frequent product changes, or small batch sizes — exactly the market reality most European manufacturers face today.

The solution

What was built

The project delivered an integrated dual-arm (bimanual) assembly robot with tactile sensors, vision, and force sensing. Key demo deliverables include tactile sensor integration, an integrated SARAFun bimanual robot system, and validated use case deployments. The system can learn from human demonstration, auto-generate assembly programs with exception handling, and design 3D-printable grippers tailored to specific parts.

Audience

Who needs this

Electronics manufacturers dealing with frequent product changeoversAutomotive Tier 1-2 suppliers assembling multi-part componentsConsumer goods companies with high product variety and short runsContract manufacturers serving multiple clients with different assembly needsMedical device assemblers needing flexible but precise production

Business applications

Who can put this to work

Electronics Assembly

mid-size

Target: Consumer electronics or mobile phone component manufacturer

If you are an electronics manufacturer dealing with frequent product changeovers on your assembly lines — this project developed a dual-arm robot that learns new assembly tasks by watching a human worker. Instead of weeks of reprogramming for each new phone model or PCB variant, a non-expert operator can teach the robot a new bimanual assembly task in less than a day. The robot even 3D-prints its own custom grippers for the specific parts.

Automotive Components

enterprise

Target: Tier 1-2 automotive supplier assembling multi-part modules

If you are an automotive supplier struggling to automate assembly of components that change with every model year — this project built a robot system with tactile sensing, vision, and force feedback that learns insertion and assembly tasks from human demonstration. The system generates its own assembly program including error handling, cutting the integration time for new tasks to less than a day instead of requiring specialist robot programmers.

Contract Manufacturing

SME

Target: Contract manufacturer handling diverse assembly jobs for multiple clients

If you are a contract manufacturer where every client brings different products and traditional robot programming eats your margins — this project created a learning-by-demonstration system for a collaborative dual-arm robot. A floor operator with no robotics expertise can teach new assembly tasks. With 3 demonstrated use cases and integrated tactile sensors, the system handles the variety that makes automation uneconomical today.

Frequently asked

Quick answers

What would it cost to implement this robot system?

The project does not disclose specific pricing or licensing costs. The base platform is FRIDA (now part of Hitachi Energy / ABB's collaborative robot line), so implementation costs would depend on ABB's commercial terms plus any licensing for the SARAFun learning software. Contact the coordinator for commercial availability details.

Can this scale to a full production line with multiple robots?

The project demonstrated an integrated bimanual robot with validated use cases. Scaling to multiple stations would require replicating the setup per workstation. Since the system is designed so non-expert users can teach tasks in less than a day, scaling the knowledge transfer across stations is significantly faster than traditional programming.

Who owns the intellectual property and can we license it?

The consortium was led by Hitachi Energy Sweden AB (formerly ABB) with 6 partners across 4 countries. As a Research and Innovation Action (RIA), IP typically stays with the partners who generated it. ABB/Hitachi Energy likely holds key exploitation rights for the robot platform and learning system. Licensing terms would need to be discussed directly with them.

How long does it really take to teach the robot a new task?

The project's explicit target was enabling a non-expert user to integrate a new bimanual assembly task in less than a day. This includes the robot observing a human demonstration, analyzing the task, generating an assembly program with exception handling, and 3D-printing custom grippers for the specific parts.

Does it work with our existing equipment or do we need to replace everything?

The system was built around ABB's FRIDA collaborative robot, designed to fit into existing workplaces without dedicated safety caging. The robot uses standard vision, force, and tactile sensors. However, the learning-by-demonstration software and 3D-printed gripper system are specific to this platform.

What types of assembly tasks can it handle?

Based on the project data, the robot was trained on tasks including insertion and folding operations. It uses bimanual (two-arm) manipulation with tactile sensing, vision, and force feedback. The 3 demonstrated use cases were validated through the SARAFun use cases deployment deliverable, though specific task details would need to be confirmed with the consortium.

Consortium

Who built it

The consortium of 6 partners across 4 countries (Germany, Greece, Spain, Sweden) is heavily research-oriented with 3 universities and 2 research organizations supporting a single industrial partner. That industrial partner is significant — Hitachi Energy Sweden AB (formerly ABB), a global leader in robotics and automation. With a 17% industry ratio and zero SMEs, this is clearly a technology-push project driven by a major corporate player with the resources to commercialize results. The fact that ABB built the base robot (FRIDA) and led the consortium suggests strong intent to bring results into their product line, though the academic-heavy partnership means commercial readiness may require further engineering.

HITACHI ENERGY SWEDEN ABCoordinator · SE
KUNGLIGA TEKNISKA HOEGSKOLANparticipant · SE
ETHNIKO KENTRO EREVNAS KAI TECHNOLOGIKIS ANAPTYXISparticipant · EL
LUNDS UNIVERSITETparticipant · SE
FUNDACION TECNALIA RESEARCH & INNOVATIONparticipant · ES
UNIVERSITAET BIELEFELDparticipant · DE

How to reach the team

Hitachi Energy Sweden AB (formerly ABB) — reach out to their robotics division or collaborative automation team in Sweden

Order a report on this consortium See HITACHI ENERGY SWEDEN AB profile →

Next steps

Talk to the team behind this work.

Want to explore how SARAFun's rapid robot teaching technology could cut your assembly changeover times? SciTransfer can connect you with the right people in the consortium. Get in touch for a one-page brief.

Order a report on this topic More in Manufacturing & Industry 4.0