Intelligent Robotics II

O1: To know the different elements of a cognitive architecture as they are usually implemented in autonomous robots.
O2: To know the particularities of learning techniques when used in robotics, paying special attention to open and continuous learning, as well as to collaborative learning, either with other robots or with humans, for problem solving.
O3: Know how to implement, even in simplified form, examples/elements of everything seen in theory (components of a cognitive architecture, learning methods).

The general contents indicated in the degree report are: Representation and modelling. Reasoning and decision-making. Learning in robotics (real time, uncertainty, adaptation to the environment). Cognitive architectures in autonomous robotics: mechanisms of motivation and attention, redescription and consolidation of knowledge, types of memory, developmental robotics. Open-ended learning.

These contents are developed in the subject through 6 units organised into 2 large blocks and presented in 21 hours of lectures distributed according to the indications given in each unit:
BLOCK 1: PLANNING
- Unit 1: Automatic Planning: Introduction. Representation in planning. Planning methods (4,5h on-site and 4,5h off-site).
- Unit 2: Uncertainty in Robotics: Introduction. Decision-making with uncertainty. Interleaving planning and execution (3h on-site and 3h off-site).
- Unit 3: Reinforcement learning: Definition. Learning the Q-function. Representation of the Q-function. Deep reinforcement learning. Sim2Real (3h on-site and 3h off-site).
BLOCK 2: COGNITIVE ARCHITECTURES
- Unit 4: General functional scheme of cognitive mechanisms: Primitive components. Basic processes related to search and optimisation. Cognitive decision processes. Domain and Context. Study of a simple example of cognitive architecture and its processes. (4,5h on-site and 4,5h off-site).
- Unit 5: Short-term and long-term memories. Types of memory. Difference between computer memory and cognitive memory. Some example implementations looking at their problems and possible solutions. Study of examples of implementations in real robots (3h on-site and 3h off-site).
- Unit 6: Lifelong Open-ended learning. Associative learning. Perceptual and performance classes. Motivation. Processes of redescription and autonomous learning of representations (3h on-site and 3h off-site).

Three labs will also be developed during the interactive classes (with 21 face-to-face hours and 87 non-face-to-face hours, distributed as indicated in each lab) to illustrate in an experimental way the concepts of the theoretical units:
- Lab 1: Application of classical planning algorithms to a robotic task. It will illustrate the concepts of units 1-2. (7,5h on-site and 31h off-site).
- Lab 2: Application of reinforcement learning algorithms to a robotic task. It will illustrate the concepts of unit 3. (3h face-to-face and 13h non-face-to-face).
- Lab 3: Application of a cognitive architecture to a robotic task. It will illustrate the concepts of units 4-5-6. (10,5h face-to-face and 43h non-face-to-face).

Basic bibliography:
- Stuart J. Russel, Peter Norvig. Artificial Intelligence: A Mordern Approach, 4th Edition, 2020. Pearson. ISBN: 978-0134610993
- Richard S. Sutton, Andrew G. Barto. Reinforcement Learning: An Introduction, 2nd Edition, 2018. ISBN: 9780262039246
Supplementary bibliography:
- Bruno Siciliano, Oussama Khatib. Springer Handbook of Robotics, 2nd Edition. Springer, 2016. ISBN: 978-3-319-32552-1.
- Robin R. Murphy. Introduction to AI Robotics, 2nd Edition, MIT Press, 2019. ISBN: 9780262038485.

Core: CB6, CB7, CB9.
General: GC1, GC2, GC3.
Transversal: CT3, CT5, CT7, CT8.
Specific: CE17, CE18.

For further information, please consult the verification report at the following link:
https://pro-assets-usc.azureedge.net/cdn/ff/cNqOjOOYIgRoeuEESpxcTyBCWBq…

The contents of the subject will be taught both in lectures and in interactive practical classes. Both types of classes will alternate throughout the semester, in such a way that the practical classes will reinforce the concepts previously taught in theory.

The lectures will be developed in the classroom by the teaching staff, supported by electronic media (electronic presentations, videos, complementary technical documents...) available on the Virtual Campus of the USC. These classes will follow the detailed contents of the subject as reflected in the annual teaching programme. The lectures will be combined with specific exercises to reinforce the concepts presented. These exercises will be solved by the students (in class or at home) and then corrected in class in a participative way.

The teaching of the labs, of an interactive nature, will be complementary activities to the theoretical lectures. They will be carried out in robotics laboratories with real robots and in computer classes with robotic simulators, under the supervision of the teaching staff. Students will autonomously follow the lab scripts available on the USC Virtual Campus. These activities will not only allow students to understand the theoretical concepts by putting them into practice, but will also enable them to acquire the necessary skills to program real robots in their professional future.

Likewise, in the tutorials, students will be attended to discuss, comment, clarify or resolve specific questions in relation to their tasks within the subject (information gathering, preparation of assessment tests, practicals, assignments...). These tutorials will be both face-to-face and virtual via email, virtual campus or the Microsoft Teams platform.

The evaluation of the subject will consist of two different parts: theory (50%) and practical work (50%). The theoretical part will be assessed by means of an exam that may consist of an analysis of a scientific bibliography related to the subject, presented orally on the day of the final exam. The practical part will be assessed on the basis of the average of the reports presented at the end of each lab, weighted according to the number of classroom hours dedicated to each of them.

Attendance to both theory and practical classes will be compulsory in order to pass the subject, except in cases of justified absence. For those students who have an exemption, the evaluation system will be the same although they will not be obliged to attend the theory classes. Resitters will not be obliged to attend the theory classes and may not attend the practical classes if they passed the practical part of the subject in a previous exam session and decide to keep that grade.

Second chance assessment and resitters: Students must make up each failed part (theory and/or labs). If one of the two parts was previously passed, the student may choose to keep the corresponding grade and only make up the failed part. In order to make up the labs, students must hand in, on the date prior to the theory exam, those activities that the teacher will ask them to do corresponding to the labs that they have not passed previously. There may be a defence of these activities with the teacher. To recover the theoretical part, there will be a final exam as in the first opportunity.

The student will receive the grade of "no-show" when he/she does not take the final exam of the theoretical part or when he/she does not present any practical.

The subject-specific competences as well as the general-basic competences have specific contents in the subject that are introduced, as indicated, both in the lectures and in the interactive classes. Subsequently, students will develop these competences in the theoretical exam and by carrying out practical work in which they will also work on the transversal competences, especially with regard to the ability to use ICT tools (CT3), the understanding of entrepreneurial culture (CT5), the ability to work in a team (CT7) and the assessment of research and innovation (CT8). The specific competences will be assessed both in the practical work carried out by the student during the course and in the theory exam.

In cases of fraudulent performance of exercises or tests, the provisions of the "Regulations on the evaluation of students' academic performance and review of grades" shall apply.

The subject has a workload of 6 ECTS divided as follows:
- 1. Classroom work (42 classroom hours)
- 1.1. Theoretical lectures: 21 hours
- 1.2. Internships: 14 hours
- 1.3. Seminars, problems, case studies and projects: 7 hours
- 2. Students' personal work (108 non-face-to-face hours)
- 2.1. Reading, review of units and theory exercises: 21 hours
- 2.2. Preparation of exercises and lab reports: 87 hours

In order to develop the objectives of the subject, students should review the basics obtained in the subject "Intelligent Robotics I".

Due to the high correlation between the concepts developed in the theory classes and the contents of the practicals, students are recommended to be consistent in the study of the subject, attending the practical sessions with the revised theoretical concepts and the solved exercises. The practical sessions will help to consolidate the theoretical concepts and their use in real situations.

The language of instruction will be English.