CoLLaboratE · Project

Robots That Learn to Work Alongside Your Assembly Workers — No Safety Fences Needed

manufacturingPilotedTRL 6

Imagine teaching a new coworker a task just by showing them how it's done — that's what this project did with robots. Instead of programming every move, workers demonstrate assembly steps and the robot learns to help out, passing tools, holding parts, and sharing the heavy lifting. The system figures out who's better at what (human or robot) and splits the work accordingly. It was tested on real factory floors across 4 countries, with robots working right next to people without protective cages.

By the numbers

Industrial pilot sites across different countries

Consortium partners

Countries represented in consortium

EUR 6,864,600

EU funding for development and validation

Total project deliverables produced

Demo deliverables including final integrated co-production cell

Industry partners in consortium

The business problem

What needed solving

Most manufacturers are stuck between two bad options: expensive full automation that breaks every time the product changes, or slow manual assembly that can't scale. Reprogramming industrial robots for each new product variant takes weeks and specialized engineers. Meanwhile, skilled workers handle complexity well but fatigue on repetitive heavy tasks, creating bottlenecks and quality issues on the assembly line.

The solution

What was built

The project delivered a final integrated co-production cell where robots learn assembly tasks from human demonstration, plus automated guided vehicles (AGVs) for material transport. A complete safety system for fenceless human-robot collaboration was built, including human tracking, touch recognition, and intention detection. All components were demonstrated at 4 industrial pilot sites, producing 47 deliverables in total.

Audience

Who needs this

Automotive assembly plants dealing with frequent model changeoversElectronics contract manufacturers with short production runsIndustrial robot integrators building collaborative solutions for clientsAerospace component manufacturers requiring flexible precision assemblyLogistics and warehouse operators looking to add collaborative AGVs

Business applications

Who can put this to work

Automotive manufacturing

enterprise

Target: Car or automotive parts manufacturers with mixed-model assembly lines

If you are an automotive manufacturer dealing with frequent model changes that make full automation too expensive and too slow to reconfigure — this project developed an integrated co-production cell where robots learn assembly tasks from human demonstration and adapt in real time. Tested across 4 pilot sites in Italy, Slovenia, Turkey, and Romania, it lets you keep your skilled workers while boosting throughput on repetitive or heavy-lifting steps.

Electronics assembly

mid-size

Target: Contract electronics manufacturers handling short production runs

If you are an electronics assembler struggling with short production cycles that don't justify traditional robot programming — this project built self-learning robots that pick up new tasks by watching a worker do them once. The system uses deep reinforcement learning to improve over time, and automated guided vehicles to move parts between stations, cutting changeover time when switching between product variants.

Industrial automation and system integration

SME

Target: Robot integrators and automation consultants serving SME manufacturers

If you are a system integrator looking for a ready-tested collaborative robotics package to offer your manufacturing clients — this project delivered a complete fenceless co-production cell with human tracking, touch recognition, and intention detection built in. With 47 deliverables including AGV deployment guides and safety strategies validated at 4 industrial sites, it gives you a technology stack to build commercial solutions on.

Frequently asked

Quick answers

What would it cost to implement this kind of collaborative robot cell?

The project itself received EUR 6,864,600 in EU funding across 16 partners over 3.5 years to develop and validate the technology. Individual cell deployment costs are not published, but the fenceless design eliminates expensive safety cage infrastructure, which is typically a major cost component in robotic assembly setups.

Can this scale to a full production line, not just a demo cell?

The system was validated at 4 different industrial pilot sites across Italy, Slovenia, Turkey, and Romania, each with different assembly challenges. The architecture uses automated guided vehicles (AGVs) and a resource allocation system that distributes tasks between humans and robots, which is designed to scale across multiple cells on a factory floor.

Who owns the intellectual property and can I license this technology?

The project was coordinated by Aristotle University of Thessaloniki with a consortium of 16 partners including 5 industry partners and 2 SMEs across 10 countries. IP ownership and licensing terms would be governed by the consortium agreement. Contact the coordinator or industrial partners directly for licensing discussions.

Does this meet safety regulations for fenceless human-robot work?

A core focus of the project was developing effective safety strategies specifically for fenceless operation within the production cell. Human tracking and human touch recognition systems were built in as fundamental safety layers. The technology was demonstrated at 4 industrial sites, suggesting compliance testing was part of the validation process.

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

The system uses learning from human demonstration — a worker shows the robot what to do rather than an engineer programming each movement. Deep reinforcement learning then improves the robot's performance over time. Based on available project data, specific teaching times per task are not published, but the entire approach was designed to dramatically shorten deployment time compared to traditional robot programming.

Can this integrate with our existing robots and factory systems?

The project developed AGVs (automated guided vehicles) alongside the collaborative robot cells and a production planning system that allocates tasks between available human and robotic resources. With 5 industry partners in the consortium contributing real factory environments, integration with existing industrial setups was a core design requirement.

Is there ongoing technical support or has the project ended?

The project officially closed in March 2022. However, the consortium includes 5 industry partners and 2 SMEs who may be commercializing elements of the technology. The project website (collaborate-project.eu) and the 47 published deliverables remain available as technical references.

Consortium

Who built it

This is a strong, industry-grounded consortium with 16 partners from 10 countries. The 31% industry ratio (5 industrial partners including 2 SMEs) means the technology was shaped by real manufacturing needs, not just academic interest. The 4 pilot sites at industrial partners in Italy, Slovenia, Turkey, and Romania provided genuine factory-floor validation across different assembly contexts. With 6 universities and 5 research organizations backing the technical work, and EUR 6,864,600 in EU funding, this had serious resources behind it. For a business buyer, the key signal is that multiple manufacturers already trusted this technology enough to test it on their own production floors.

ARISTOTELIO PANEPISTIMIO THESSALONIKISCoordinator · EL
ETHNIKO KENTRO EREVNAS KAI TECHNOLOGIKIS ANAPTYXISparticipant · EL
CENTRO RICERCHE FIAT SCPAparticipant · IT
INSTITUT JOZEF STEFANparticipant · SI
UNIVERSIDAD DE BURGOSparticipant · ES
ASTI MOBILE ROBOTICS SAparticipant · ES
ASSOCIATION POUR LA RECHERCHE ET LE DEVELOPPEMENT DES METHODES ET PROCESSUS INDUSTRIELSparticipant · FR
ARCELIK A.S.participant · TR
FONDATION DE L'INSTITUT DE RECHERCHE IDIAPparticipant · CH
ROMAERO SA*participant · RO
BLUE OCEAN ROBOTICS APSparticipant · DK
UNIVERSITA DEGLI STUDI DI GENOVAparticipant · IT
KATHOLIEKE UNIVERSITEIT LEUVENparticipant · BE
ECOLE NATIONALE SUPERIEURE DES MINES DE PARISthirdparty · FR
PANEPISTIMIO PATRONparticipant · EL

How to reach the team

Aristotle University of Thessaloniki (Greece) coordinated the project. Industrial partners across Italy, Slovenia, Turkey, and Romania hosted pilot sites and may offer the most direct path to technology access.

Order a report on this consortium See ARISTOTELIO PANEPISTIMIO THESSALONIKIS profile →

Next steps

Talk to the team behind this work.

Want an introduction to the CoLLaboratE team or a detailed technology brief for your assembly line? Contact SciTransfer — we connect manufacturers with EU research teams that have already solved your problem.

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