What would it cost to license or adopt this robot motion technology?+
The project produced open source software for learning predictive models in sensor space, which can be accessed freely. For the full integrated systems (humanoid, exoskeleton, quadruped), licensing terms would need to be discussed directly with the consortium led by CNRS. As a publicly funded RIA project, results are generally accessible.
Can this scale to a full production line, not just a demo?+
The aircraft assembly demonstrator was built at 1:1 scale — a full-size replica of a real assembly environment, not a scaled-down lab model. The consortium includes 5 industry partners (42% of the 12-partner team) who designed the demo specifications, suggesting the technology was built with real production requirements in mind. Scaling to continuous production would require further engineering.
Who owns the IP and can I use this technology?+
CNRS (France's national research center) coordinated the project across 12 partners in 6 countries. IP is shared among consortium members according to their EU grant agreement. At least one deliverable — the predictive models software — is explicitly open source. For proprietary components, licensing negotiations would go through the relevant consortium partner.
How mature is this — lab experiment or ready to deploy?+
The project ran 3 real-world demonstrators: aircraft assembly at 1:1 scale, exoskeleton walking in a rehabilitation center, and quadruped inspection on a real construction site. These were tested in relevant operational environments with end-user specifications, putting it beyond lab stage. However, as a research project (RIA), additional product engineering would be needed before commercial deployment.
Can this integrate with our existing robot fleet or control systems?+
The approach is built on optimal-control theory and works with any robot that has arms and legs — the objective explicitly targets 'any combination of arms and legs.' The system uses standard sensor modalities (vision, inertial, haptic) for feedback control. With 29 deliverables produced, integration documentation should be available through the consortium.
What sensors and hardware does this require?+
The system exploits vision, inertial, and haptic sensor modalities for feedback control. It was demonstrated on 3 different robot types (humanoid, exoskeleton, quadruped), showing it adapts to different hardware platforms. The open source software component focuses on learning predictive models in sensor space and computing reduced models using sensory features.
Is there regulatory approval for the medical exoskeleton application?+
The exoskeleton was demonstrated in a rehabilitation center under medical surveillance with a paraplegic patient, suggesting clinical oversight was in place for testing. However, this was a research demonstration — full medical device certification (e.g., CE marking) would be a separate process required before commercial use with patients.
