Spatial Intelligence & Embodied Engineering

Timetable

60 hour course with lectures and laboratories (3h class weekly) at the Faculty of Physics and Astronomy.

Located in: UAM Morasko Campus

Address: Uniwersytetu Poznańskiego 2, 61-614 Poznań

Module aim (aims)

The aim of this module is to equip students with the engineering capability to build Digital Twin Simulators using Generative AI tools, and then use those simulators to train Embodied AI Agents. We focus on the Sim-phase of the "Sim-to-Real" pipeline: leveraging simulated physics of 3D/VR IDEs to pre-train brains that could operate then in the real world. Students will move from "Manual Modeling" to "Agentic World Building", using AI to generate assets and environments. The simulated worlds will be published to VRChat, enabling "Social Validation" and massive human-in-the-loop testing.

Crucially, this module serves as a practicum in "Human/Team-in-the-loop Agentic Management". Students will not just build; they will orchestrate AI teams, preparing for high-end Executive Engineering roles where the "Team" includes both humans and autonomous agents.

By the end of the course, students will have built and trained agents in four distinct physics domains: Evolutionary Swarms, Ground Vehicles, Aerial Drones, and Fixed-Wing Aircraft.

Pre-requisites in terms of knowledge, skills and social competences (where relevant)

Module 1 (Foundations) is designed as an accelerated "Agentic Boot-Camp", covering the critical Generative AI, Orchestration, and Project Management skills required for the 4 Case Studies. This ensures accessibility for students entering directly into course without prior Agentic training.

Syllabus

1. Foundations: Generative 3D Modeling & Human/Team-in-the-Loop Management
2. Foundations: Reinforcement Learning & Physics Engines
3. Foundations: Sim-to-Real, 3D/VR Deployment
4. CS1: Evolutionary Ecosystem - AI-Assisted (AIA) Environment Building
5. CS1: Evolutionary Ecosystem - Multi-Agent RL (MARL) Training
6. CS2: Autonomous Vehicles - AIA Environment Building
7. CS2: Autonomous Vehicles - MARL Training
8. CS3: Drone Flight - AIA Environment Building
9. CS3: Drone Flight - MARL Training
10. CS4: Plane Flight - AIA Environment Building
11. CS4: Plane Flight - MARL Training
12. Final Defense: The Embodied Turing Test (in VRChat)

Assessment (Course Portfolio) The Semester Project is mandatory. Students are to build a Digital Twin Simulator for a real-world system and train Embodied AI Agents within this environment.

Reading list

Scripts, presentations, webinars, and Agentic Design Patterns delivered by the lecturer. Access to the Spatial Intelligence Stack.