ai

Advanced Topics in Artificial Intelligence: Planning and Learning for Intelligent Robotics

COMP 4620/8620 – Semester 2 2026#

Lecture: Tuesday 09.00-11.00am Canberra time, Rm G08 Psychology Bldg 39

Notes 27/07/2026: There has been issues with the ANU Permission Code and Enrollment System for this class. The ANU and SoCo Education Team are aware and working to resove these issues. In the mean time, even if your enrollment is still pending, you are very welcome (in fact, highly encouraged) to come to class. Details about the lecture time and location are in this website.

How can a robot develop strtegies to accomplish its tasks or missions in our world? This is the question we will explore in this class. Specifically, we will focus on sequential decision-making in robotics.

Robots have the potential to make society safer and more efficient, from reducing the need for humans to perform dangerous jobs to enhancing services in remote areas to helping busy parents with everyday chores. However, their wide-spread use have often been hindered by the ubiquity of uncertainty. In this class, we will explore and discuss some of the concepts, approaches, and tehniques that would enable robots to embrace and work with uncertainty, rather than avoiding them.

In this class, we will view robots in a broad sense: Intelligent agents that operate in the physical or simulated physical world and will discuss concepts, basic approaches, and recent scalable approaches of:

  • Planning in continuous and hybrid spaces (in contrast to the discrete space planning in COMP3620/6320)
  • Sequential decision-making when the effects of actions are uncertain (i.e., Markov Decision Processes)
  • Sequential decision-making when the effects of actions are uncertain and the world is only partially observable (i.e., Partially Observable Markov Decision Processes)
  • Sequential decision-making when models are unavailable a priori (i.e., Reinforcement Learning and integrated planning and learning)

Although we will focus on robotics applications, many of the approaches we will discuss are general and can be applied to various domain and occasionally, if time permits, we will discuss applications in other domains, such as in cyber domain.

Activities#

This class is an in-person class. Hence, we do expect that you come to lectures and tutorials in-person to gain the most from the class. We will try to record the lectures, but we do NOT guarantee that there will be a recording for every class.

Lecture: Tuesday 09.00-11.00am Canberra time, Rm G08 Psychology Bldg 39

Tutorial + Drop Ins: Will start in week-2, please see timetable for the exact time and location

We will not follow a particular book, we will use chapters from a few different books and research papers. The exact reading materials will be provided prior to lecture time.

All class materials will be available in the wattle page of this class.

Assessments#

This class will have 3 assessments:

  • Assignment 1 (conceptual questions + problem solving + programming + demo): 25%
  • Assignment 2 (conceptual questions + problem solving + programming + demo): 35%
  • Final Exam (in-person, all materials): 40%

Teaching Staff#

Convenor + lecturer: Hanna Kurniawati

bars magnifying-glass xmark