Abu dhabi: A new physical AI-framework called Tactile Skills is revolutionizing robotic automation by enabling robots to quickly learn and perform complex, delicate tasks that have historically been challenging.
According to Emirates News Agency, Sami Haddadin, Vice President for Research at Mohamed bin Zayed University for Artificial Intelligence (MBZUAI), collaborated with his former PhD student Lars Johannsmeier, along with Yanan Li from the University of Sussex and Etienne Burdet from Imperial College, to develop this innovative framework. Their work was published in Nature Machine Intelligence on June 23, highlighting a method inspired by the human neural system and vocational training.
The Tactile Skills framework employs a structured taxonomy, akin to providing robots with a specialized curriculum, facilitating rapid mastery of new tasks. “Our framework bridges the gap between human expertise and robotic capability,” stated Haddadin, emphasizing its transformative impact on robotic precision and adaptability.
The approach has undergone extensive testing across 28 industrial tasks, achieving nearly 100% success in complex operations such as plug insertion and precision cutting. Notably, the robots demonstrated remarkable adaptability to changes in object positioning and environmental conditions, maintaining high performance with minimal error.
Unlike traditional machine-learning methods that rely on large datasets and trial-and-error, this framework integrates expert process knowledge with tactile control components, simplifying the learning process and reducing energy consumption. This innovation enhances performance significantly compared to deep-learning-based approaches.
A significant milestone was the successful assembly of a complex industrial device for bottle-filling plants, showcasing the framework’s real-world applicability. By structuring robotic learning, this method empowers operators without deep robotics expertise to deploy robots efficiently, cutting setup times and costs.
“This research marks a pivotal moment in automation, transforming robots into adaptable, skilled assistants,” Haddadin noted. It opens possibilities for automating complex, tactile tasks across industries, enhancing efficiency, safety, and scalability. Ultimately, this breakthrough extends robotic capabilities, making versatile skills accessible in industrial settings and potentially in everyday life.