https://www.pw.edu.pl/engpw
Research University - Warsaw University of Technology

Artificial Intelligence and Robotics

Global and local challenges

Artificial Intelligence is a branch of computer science which is focused on building models from data gathered from an environment in order to predict future states of the environment and to make decisions which are based on these predictions. A particular interest in AI is on the development of models to represent the knowledge, and of effective methods to learn models from data. Diversity of models include: decision rules, association rules, decision trees, neural networks, Bayesian models, support vector machines, fuzzy logic, semantic networks, etc. Decision support methods which fall into the scope AI include heuristic search methods, genetic and evolutionary algorithms, and metaheuristics. Typical fields of AI applications are machine vision, natural language processing, computer aided design, automatic control and robotics. AI based systems are applied in various businesses such as medicine (in particular in medical imaging), finance (e.g. automated trading), marketing (customer profiling and personalized advertising, chatbots), industry (smart grid, industry 4.0), transportation (intelligent and autonomous vehicles), entertainment (computer games), and many others. Rapid development of electronics, sensors, programmable hardware, broadband wireless communication, and computer systems gave an opportunity to acquire and process an enormous amount of data. In the same time, there have been achieved several milestones of AI development, which made many political institutions to recognize AI as a crucial factor for the development of economy and national security. For these reasons both EU and Poland are planning to make AI the priority area of scientific development. Among various fields of AI applications, robotics is particularly demanding, since the robot construction demands effective solving of various tasks, which include awareness of the state of the robot and its environment, planning of robot’s further actions, and communication with humans and other robots.

For these reasons, when AI is applied in robotics, the term embodied intelligence is used to define such a combination. Further on, we us the acronym AIR to refer to the research and development on the crossing between Artificial Intelligence and Robotics.

An important trend in the modern robotics is the development of cobots (co-robots, collaborative robots) which are the robot to support the professional human in his/her activity in the factory, or in the office. Cobots are flexible, human-safe, and as such can be easily adopted (trained) to perform various and changing tasks. It is predicted that the market of co-bots in years 2014-2020 will increase by 50% and the income will exceed $1 billion.

Aging of modern societies is an increasing problem which makes it necessary to develop care robots, nursing robots and companion robots. First robots of that kind were produced in Japan. An example is Reysone by Panasonic, which is a bed which can transform into the autonomous electric wheelchair. RIKEN Institute has developed an anthropomorphic robot Robobear which is capable of performing tasks such as lifting a patient from a bed into a wheelchair or providing assistance to a patient who is able to stand up but requires help to do so. The Israeli start-up "Intuition Robotics" has acquired over $14 million from the investors, including Toyota Research Institute, to develop the companion robot ElliQ dedicated for elderly people.

The care robots are also in the scope of IBM. The company concentrates on the development of sensors and software to facilitate humanlike communication with elderly persons (including gesture and facial expression) and to effectively monitor the health condition. The examples of robots incude IBM MERA (ang. Multi-Purpose Eldercare Robot Assistant) and OBI manipulators to feed persons who are unable to use cutlery.

Scientific excellence

At WUT, research and development on the crossing between Artificial Intelligence and Robotics (AIR for further reference) is in the scope of scientific interests of 98 researchers employed by the University. The research profile of the core group in AIR can be characterized by the overall scientific output is set at the level of 1,009 publications, with 4.483 citation and with the Field- Weighted Citation Impact equal to 1.03, according to SciVal statistics for years 2013-2017.

The main research activities of the AIR core group are:

  • in the field of AI: knowledge representation, machine learning, machine vision, pattern recognition, natural language processing,
  • in the field of robotics: mobile robotics, robotic manipulation, automatic control for robotics,
  • in the field of applied mathematics: modelling and simulation of systems, and numerical optimization.

WUT has been actively participating in the national debate on the goals and means to develop AI – we cooperate with other Polish academic partners under the umbrella of PP-RAI (Polish Platform for AI Development) alliance. We also participate in expert groups formed by the Government to formulate strategies aimed at developing AI in Poland. Researchers of the core AIR group are members of scientific boards of many scientific institutions, including Polish Academy of Sciences and PIAP (Institute of Automatic Control and Measurements).

WUT is active in the organization of national and international conferences. For example, in years 2013-2017, WUT was the organizer of the following conferences within the scope of AIR: International Conference on Pattern Recognition and Machine Intelligence (PReMI 2015), International Symposium on Methodologies for Intelligent

Systems (ISMIS 2017), EUROPT Workshop on Advances in Continuous Optimization (EUROPT 2016), Federated Conference on Computer Science and Information Systems (FedCSIS).

Research plans

WUT plans to work on its own companion robot dedicated for caring for elderly persons. The examples of functionalities of that robot are:

  • health and safety monitoring – detection of abnormal states, like a downfall, fainting, stroke,
  • finding, taking and handing of things, e.g. books from shelves,
  • controlling the routine of the day, like advising to take medicines,
  • serving food and dishwashing.

The examples of the companion robot’s abilities include:

  • communication and co-operation with other companion robots,
  • detection of its position and identification of objects in the environment, together with their location,
  • communication in natural language and/or with gestures,
  • manipulation of objects according to the human commands.

The robot should be equipped with an artificial "mind" to be able to understand the environment and the human, to predict changes, to plan future actions, to infer about the actions, consequences, etc. Effective implementations of such functionalities are demanding scientific problems which can be solved by teams of researchers rather than individuals. This team work will strengthen the cooperation between researchers at WUT and it will build a natural environment to accelerate the development of both senior and junior scientists. On top of that, it will attract new students.

Education

The following curricula are directly linked with AIR:

  • course in "Computer Science and Information Systems", track "AI methods", BSc and MSc degrees (Faculty of Mathematics and Information Science),
  • courses in "Computer Science”, BSc and MSc degrees (Faculty of Electronics and Information Technology),
  • courses in "Automatic Control and Robotics", BSc and MSc degrees (Faculty of Electronics and Information Technology, Faculty of Electrical Engineering, Faculty of Power and Aeronautical Engineering, Faculty of Mechatronics).

In October 2019, the Faculty of Electronics and Information Technology will launch a new BSc course in "Computer Science" which will include the track "AI methods". The course will cover the AI in a broad sense, with a special accent on applications in robotics. Major innovation of the new course is the implementation of the projectbased learning paradigm. The students will work in teams on two projects, for each of them is foreseen one semester.

The projects will be focused both on real-life and business cases, and on scientific research problems, depending on students’ choice. In October 2020, the Faculty of Electronics and Information Technology will launch a new MSc course in "Computer Science" which will be entirely related to AI issues and its applications. The course is intended to provide students with a strong theoretical background of AI and to make them contribute to the current scientific development at WUT.

An important role in learning is played by the student research group. It is planned to intensify their activities by granting them research projects which will correlate with works on the companion robot development. It is planned to cooperate with student research groups dedicated to artificial intelligence, robotics and automatic control, in particular with "BIONIK", "GOLEM", "RAR", and "ROBOMATIC", which are active at various Faculties.

The AI cooperation platform PP-RAI is planning to create a federation of doctoral schools of leading academic centers in AI in Poland and WUT plans to join that federation. The following activities are planned within that cooperation:

  • exchange of PhD students – scientific stays,
  • master’s degree courses run by visiting scientists included in the federations’ mobility scheme,
  • a periodical conference dedicated for PhD students.

Human capital

Artificial Intelligence and Robotics area includes over 100 researchers at WUT specialised in computer science, automatic control and robotics, decision sciences, and mathematical modelin.