Machine Learning of Intelligent Robotic Systems

The aim of the course is to gain student's ability for development and implementation of intelligent mobile robots which are capable to realize robotic tasks in advanced manufacturing environment, through hardware-software integration, without explicit human-operator control, and in accordance with the various paradigms of artificial intelligence. Considering the fact that production technologies of the 21st century involve this hardware-software integration of intelligent systems, especially mobile robots as well as automated agents, the course has its goal to educate students at master level through theoretical and practical aspects for complex development of modern robotized systems and processes, their modelling, implementation within the Industry 4.0 based on advanced algorithms and methods in domain of artificial intelligence.

After the course students will be able to: • Implement information and communication technologies within the intelligent robotic systems. • Select methods based on the application of various artificial intelligence techniques (artificial neural networks, fuzzy logic and hybrid control, etc.) as well as biologically inspired algorithms in finding the optimal solution within the process of development and application of machine learning in intelligent robotic systems (by using software packages Matlab and BPnet). • Understand the interaction between the software and hardware subsystems of intelligent mobile robot in decision-making, during exploration of the manufacturing environment, through reconfiguration its physisal structure and intelligent behaviour programming in Matlab environment. • Develop the ability for team work.

New teaching contents are: 1. Introduction to knowledge and machine learning-based intelligent systems. Machine learning models; deduction, induction and analogy. Machine learning as a basis of intelligent systems and processes. 2. Evolutiveness and intelligent systems based on multi-agent engineering methodology. 3. Intelligent mobile robots; target cognitive capabilities of mobile robots including perception processing, collision avoidance, anticipation, path planning, complex motor coordination, reasoning about other agents, etc. 4. Mobile robot localization and navigation (pose estimation) as well as characteristic objects detection in robotic exploration within the manufacturing environment. Kalman filter. 5. Artificial neural networks: multilayer feedforward neural networks, radial basis neural networks. 6. Fuzzy logic and hybrid control. 7. Neuro-fuzzy controllers. 8. Intelligent control of the mobile robot systems. 9. Biological inspired algorithms in the optimization process of intelligent robotic systems.

Practical teaching: 1. Auditory exercise: Architecture of software for machine learning of intelligent systems. 2. Laboratory work - exercise 1: Intelligent behaviour of manufacturing system agents based on empirical control algorithm. 3. Laboratory work - exercise 2: Subsumption architecture of competence levels of the intelligent robotic system (design of intelligent mobile robot behaviour in interaction with detected objects - programming in Matlab environment). 4. Laboratory work - exercise 3: Simultaneous localization and mapping (SLAM) for the mobile robot navigation - programming in Matlab environment. 5. Laboratory work - exercise 4: Communicative and interactive competence of robots in working environment. 6. Laboratory work - exercise 5: Intelligent control of a mobile robot. 7. Project activities: indoor transport of parts and/or materials; intelligent control of a mobile robot according to simultaneous localization and mapping of objects in the manufacturing environment,etc.

-

/1./ Z. Miljković, N. Slavković, M.M. Petrović, (2021) Robots, sensors and additional equipment within the Laboratory for ROBOTICS & AI: • Five-axis micro-industrial and educational robot Mitsubishi MOVEMASTER EX RV-M1; • Six-axis industrial robot ILR LOLA50; • Five-axis industrial robot "GOŠKO"; • DELTA robot with 3+1 DoF; • Khepera II – KheIIBase mobile robot with Khepera Gripper Turret, camera CMUcam VISION TURRET–KheCMUCam and infrared sensors; • Lego® Mindstorms NXT and Lego® Mindstorms EV3 mobile robots with various sensors; • "Buggy” mobile robot with infrared sensors and control module "Clicker2” based on micro-controller ARM STM32F407VGT6; • Recognition system for intelligent mobile robot control based on stereo „visual servoing". /2./ Z. Miljković, M.M. Petrović, (2021) INTELLIGENT MANUFACTURING SYSTEMS – with excerpts from robotics and artificial intelligence, University of Belgrade - Faculty of Mechanical Engineering, Textbook in Serbian (XXVIII+409 p.), (1st Edition: ISBN 978-86-6060-071-6). /3./ Z. Miljković, M.M. Petrović, (2022) Handouts for each lecture /In Serbian/, University of Belgrade - Faculty of Mechanical Engineering. /4./ Z. Miljković, R. Jovanović, M.M. Petrović, (2022) Instructions for doing students' project tasks and laboratory exercises /In Serbian/, University of Belgrade - Faculty of Mechanical Engineering. /5./ R. Jovanović, (2022) Intelligent control systems, Handouts /In Serbian/, University of Belgrade - Faculty of Mechanical Engineering. /6./ R. Jovanović, (2022) Fuzzy control systems, Handouts /In Serbian/, University of Belgrade - Faculty of Mechanical Engineering.

Total assigned hours: 90

New material: 25
Elaboration and examples (recapitulation): 5

Auditory exercises: 5
Laboratory exercises: 30
Calculation tasks: 0
Seminar paper: 0
Project: 10
Consultations: 0
Discussion/workshop: 0
Research study work: 0

Review and grading of calculation tasks: 0
Review and grading of lab reports: 0
Review and grading of seminar papers: 0
Review and grading of the project: 6
Test: 2
Test: 2
Final exam: 5

Activity during lectures: 5
Test/test: 20
Laboratory practice: 10
Calculation tasks: 0
Seminar paper: 0
Project: 35
Final exam: 30
Requirement for taking the exam (required number of points): 30

Z. Miljković, M.M. Petrović, (2021) INTELLIGENT MANUFACTURING SYSTEMS – with excerpts from robotics and artificial intelligence, Textbook in Serbian (XXVIII+409 p.), UB-FME (ISBN 978-86-6060-071-6).; E. Alpaydin, (2010) INTRODUCTION TO MACHINE LEARNING, 2nd Edition (584 p.), The MIT Press (ISBN-13: 978-0262012430), Cambridge, England.; R.R. Murphy, (2000) INTRODUCTION TO AI ROBOTICS, 2nd Edition (486 p.), Bradford Books (ISBN-13: 978-0262133838), Cambridge, England.; Miljković,Z., Aleksendrić,D., (2018, 2nd Edition) ARTIFICIAL NEURAL NETWORKS – solved examples with short theory background, Textbook in Serbian (VI+225 p.), UB-FME (ISBN 978-86-7083-961-8).; R. Jovanović, (2021) Matlab and Simulink in automatic control, Textbook in Serbian (319 p.), University of Belgrade - Faculty of Mechanical Engineering (ISBN 978-86-6060-091-4 ).