Artificial Intelligence for Robotics

Course Type: Lab Course (Master’s Level)
Language: English
Time & Place: Wednesdays, 16:30 - 18:00 | AI Center Building Lecturers Daniel Swoboda, M.Sc., Ulzhalgas Rakhman, M.Sc.

Study programs

Overview

How do we make robots solve complex tasks autonomously? This lab course explores the algorithmic foundations of intelligent robotics. Students will build a full autonomy stack in the MuJoCo simulation environment, treating navigation, manipulation, and task planning as fundamental search and learning problems.

This course bridges the gap between high-level Artificial Intelligence approaches and robotic control. We focus on the “brain” of the robot: the algorithms that allow it to perceive, reason, and act in complex environments.

The central philosophy of the course is “From Search to Learning.” We will explore how fundamental capabilities—from pathfinding to object manipulation—can be modeled as explicit search problems, and how Deep Reinforcement Learning (RL) can be used to master these tasks through experience.

Content

The course is divided into two phases:

Phase I: The AI Toolbox (Exercises)

In the first half of the semester, you will work through a series of Python-based exercises in the MuJoCo simulator to build your foundational toolkit:

Phase II: Research Implementation (Project)

In the second half, you will work in groups to implement a simplified version of a state-of-the-art research paper (e.g., from ICAPS, RSS, or NeurIPS). This is your chance to apply what you have learned to build a novel autonomy pipeline.

Prerequisites

Bachelor degree in CS or equivalent and basic AI and ML courses, with some previous exposure to deep learning and reinforcement learning. We further recommend you have a background in:

Registration & Logistics

Examples

Examples of TAMP in Practice:

The HelloRobot Stretch 3 in action: