ACIT4840 Hands-on Machine Learning for Robotics and Control Emneplan

This course provides a foundation in Artificial Intelligence (AI) and focuses on a hands-on approach to the main Machine Learning (ML) methods used in data science and engineering, with special focus on robotics and control applications such as vision, navigation, task learning, fault diagnostics, condition monitoring.

The course aims to balance a good theoretical foundation with practical applications of ML to a selection of robotics and control related problems. Both Supervised, Unsupervised, and Reinforcement Learning are covered. Some of the main methods and algorithms for Regression, Classification, and Clustering are included. The principles of Artificial Neural Networks (ANN) and Deep Learning (DL) are covered in some detail, and some of the most commonly used NNs are used in problem solving. After covering the fundamentals of Reinforcement Learning (RL), the main RL methods are applied to example robotics and control problem solving. Genetic Algorithms and generative AI are briefly introduced. The course provides a foundation and practical skills for ML-based model development and problem solving that enables further knowledge and skill development. The curriculum shall be regularly updated in accordance with developments in this rapidly evolving field.

The course comprises two parts. The first part is a series of lecture seminars where after the presentation of each topic, the students work on hands-on exercises in class. The second part of the course is a practical project in groups.

Language of Instruction: English

Robotics and Control courses:

• ACIT4810 - Advanced Methods in Modelling, Simulation, and Control
• ACIT4820 - Applied Robotics and Autonomous Systems

No formal requirements over and above the admission requirements.

A student who has completed this course should have the following learning outcomes defined in terms of knowledge, skills and general competenc.

Knowledge:

The student:

• is familiar with the main principles in AI and has a practical understanding of the development and use of AI and ML
• has an understanding of the current application areas of AI and in particular for solving R&C problems
• has the theoretical and practical skills required to build simple ML models
• is familiar with Supervised-, Unsupervised-, and Reinforcement Learning ML methods.

Skills:

The student:

• can apply a variety of state-of-the-art ML methods in different robotics and control applications such as: machine vision, perception, inverse kinematics, navigation, reasoning, learning, fault detection and diagnostics, process control, condition monitoring, and many more.
• can evaluate the technical quality and practical value of various types of ML, and AI more general, for problem solving in robotics and control.

General competence:

The student:

• has both theoretical and practical understanding of ML methods
• can apply ML, and AI in general, to engineering problems.
• can discuss the relevance, strengths, and limitations of the different ML methods, and can mutually compare them to choose the appropriate method for the problem at hand.
• is able to solve real-world problems using ML and AI.
• can reflect on the practical, social, and ethical implications of AI in our society.

The first part of the course (nine weeks) comprises a series of whole-day lecture seminars. Students are expected to play an active role. Lecture seminars start with a lecture that introduces the topic and are followed by hands-on exercises in class.

The second part of the course (nine weeks) is a Robotics and Control project in groups of 1-3 students.

The course is completed by the students submitting a report and giving a presentation of their work.

The following required coursework must be approved before the student can take the exam:

One individual assignment consisting of 2-3 machine learning exercises using R.

The Exam has two parts:

1. Project report, completed individually or in groups of 2-3 students. The total length of the report should be between 5000-15000 words, with each student contributing with 5000-7500 words (80% of the final grade)

2. Oral project presentation (30 minutes if a group; 20 minutes if individual) (20% of the final grade)

Both the project report and oral presentation must receive a pass (E grade or above) in order to pass the course.

The oral exam grade cannot be appealed.

The project report grade can be appealed.

New/postponed exam

In case of failed exam or legal absence, the student may apply for a new or postponed exam. New or postponed exams are offered within a reasonable time span following the regular exam. The student is responsible for registering for a new/postponed exam within the time limits set by OsloMet. The Regulations for new or postponed examinations are available in Regulations relating to studies and examinations at OsloMet.

All aids are permitted, provided the rules for plagiarism and source referencing are complied with.

Grade scale A-F.

Exam part 1 (project report): one internal examiner

Exam part 2 (oral presentation): two internal examiners

External examiners are used periodically.

Professor Evi Zouganeli