Imagine a robot that doesn't just follow a script, but learns like a human apprentice. Instead of crashing when something changes, it can imagine a new way to solve the problem or even reverse its steps to fix a mistake. It uses a human's guidance to get better at its job over time, making it much more flexible in a busy workspace.
Robots currently fail in unpredictable environments because they cannot transfer knowledge between different tasks. This leads to costly downtime and rigid production lines that require manual reprogramming for every change.
A learning system that allows robots to refine skills via human feedback and experience. It includes a method for robots to detect faults and recover from errors in dynamic settings.
If you are a vehicle assembly plant dealing with unpredictable parts placement on a line — this project developed a learning system that allows robots to adapt to new domains. This means robots can recover from mistakes without stopping the entire production line.
If you are a custom electronics manufacturer dealing with frequent product design changes — this project developed a way for robots to refine skills through human feedback. This reduces the time spent reprogramming robots for every new product version.
If you are a smart warehouse operator dealing with varying object shapes and positions — this project developed a cognitive ability for robots to understand tasks across different environments. This improves the reliability of automated sorting and packing.
Based on available project data, no pricing or implementation cost information is provided.
The project aims for outcomes that are scalable across sectors and will be demonstrated in two industrial use cases to ensure suitability for production scenarios.
Based on available project data, there are no specific details regarding IP ownership or licensing terms.
The system uses human feedback to guide learning and is designed using social science methods to ensure workers perceive the robots as safe and reliable.
The project runs from 2024-01-01 to 2027-12-31, suggesting that results will be finalized by the end of 2027.
The consortium is well-balanced for technology transfer, consisting of 13 partners across 8 countries. With a 23% industry ratio (3 industrial partners) and a strong academic backbone (4 universities and 4 research centers), the project blends deep theoretical AI research with practical industrial validation.
Contact Università degli Studi di Trento
Contact us to match with the INVERSE consortium for early adoption of adaptive robotics.
