The Mechatronics Centre and WISENET Centre at UiA have joined forces in a new multidisciplinary frontier research project that combines Robotics and Explainable Artificial Intelligence. They have received NOK 16 million from the Research Council of Norway to develop robots that can learn and cooperate both between themselves and with human operators in industrial environments.
“We will design new generation decentralized controllers for multiple collaborative robots with the aid of deep learning and advanced control algorithms”, says Jing Zhou, Professor in Mechatronics at the Department of Engineering Sciences.
It means that the collaborative robots are not simply pre-programmed to perform specific tasks, but that they are capable of learning over time and perform new operations based on that.
The project is called Collective Efficient Deep Learning and Networked Control for Multiple Collaborative Robot Systems (DEEPCOBOT).
Professor Jing Zhou is the project manager. She will lead and coordinate the research project DEEPCOBOT, a collaboration between UiA and the partners Mechatronics Innovasjon Lab (MIL), Omron Electronics Norway, ABB Norway, The University of California San Diego (USA), KTH Royal Institute of Technology (SVE) og The University of Navarra (SPA).
The Mechatronics Centre and the WISENET Centre at the Faculty of Engineering and Sciences, will work together on the project. The Mechatronics Centre is a Top priority research Centre at UiA, where Professor Jing Zhou is the research Director, while the WISENET Centre has achieved recognition by the national NFR TOPPFORSK Programme and is directed by Professor Baltasar Beferull-Lozano.
There is rising demand for robots to solve complex tasks in industrial companies in southern Norway.
“The industry has an increasing demand of automation in operation, especially the demand of a safer and more efficient collaboration between multiple Cobots and human operators to integrate the best of human abilities and robotic automation”, Zhou says.
The new robots must be capable of cooperating with each other and with human operators.
“Through the process of deep learning and advanced control algorithms, the Cobots can interact both between themselves and with human operators in order to collectively learn from each other's experiences and perform cooperatively different complex tasks. This will contribute to the safe human-robot collaboration in industrial environments and the achievement of higher productivity and greater efficiency”, Zhou says.
Current Deep Learning methods do not provide the necessary adaptation and safety in complex scenarios while being time-efficient, preventing their application in real-time robotic applications.
“In order to be able to design this new generation data-driven control systems, we will advance the forefront of several areas, including Online Deep Learning for distributed teams of collaborative robots and human operators interacting with each other”, says Professor Beferull-Lozano, from the Department of ICT (by courtesy, member of the Department of Engineering Sciences).
Current solutions do not consider the interchange of information between multiple collaborative robots about previous actions and experiences for improving and accelerating the learning.
“In DEEPCOBOT, the learning of each robot will be based not only on its own experiences, but also based on other robots’ experiences, requiring a novel distributed sensor signal processing and information aggregation across the multiple wirelessly inter-connected robots”, Beferull-Lozano says.
The project period for DEEPCOBOTS is from 2020 to 2025. The plan is for the project to have three PhD candidates and one postdoctoral fellow.
The project is funded by the Norwegian Research Council's ICT and digital innovation programme, IKTPLUSS, within the section of Transformative Research. The Programme is called “Ubiquitous Data and Services – Researcher Project”. The Programme funds long-term projects that generate new knowledge and technology that promote productivity and efficiency.
UiA also has the CAREWELL and INDURB projects, which are funded through the Research Council’s IKTPLUSS Programme.