Imagine a robot that doesn't just follow a script but actually understands what it's doing and why. Most AI today just guesses based on patterns, which leads to silly mistakes or 'hallucinations.' This work gives robots a sense of awareness so they can handle unexpected surprises without crashing or failing.
Autonomous robots lack true understanding and self-awareness, leading to reliability risks and 'hallucinations' when they encounter unexpected situations in open environments.
A hybrid cognitive architecture and a physics-aware modeling module. These include formal theories of understanding and awareness translated into reusable software assets.
If you are a smart factory operator dealing with rigid assembly lines that break when a part is slightly misplaced — this project developed a cognitive architecture that augments the flexibility of manufacturing robots. This allows machines to adapt to changes in the environment without manual reprogramming.
If you are a drone service provider dealing with drone crashes during complex inspections — this project developed a physics-aware modeling module. This increases the resilience and group cohesion of drone teams operating in open-ended environments.
If you are a social robot manufacturer dealing with machines that misunderstand human intent — this project developed a theory of understanding and awareness. This improves human alignment, making social robots more predictable and trustworthy.
Based on available project data, no pricing or cost information is provided as this is a research project funded by the EU.
The project tests its theories across 3 real robot demonstrations, including manufacturing and drone teams, suggesting a path toward industrial scaling.
Based on available project data, specific licensing terms are not mentioned, though it involves a consortium of 7 partners including 1 SME.
It acts as a hybrid cognitive architecture designed to fix the 'childish errors' and hallucinations common in standard machine learning approaches.
The project runs from 2022-10-01 to 2026-09-30, with a proof-of-concept physics-aware module delivered in month 24.
The consortium is heavily research-oriented, consisting of 7 partners from 5 countries. With 4 universities and 2 research centers, the academic weight is high (86%), while industry representation is low at 14% (1 SME). This suggests the output is currently high-level theory and software assets rather than a commercial product.
Contact Universidad Politecnica de Madrid
Contact us to explore licensing the cognitive architecture assets.
