Using Large Language Models while learning

Large Language Models can be great assistants, but recent research shows that their impact depends on where you stand on the learning curve: as a novice, offloading your thinking to an LLM prevents the very practice that builds the skill, and you then perform worse without it; as an expert, delegating the routine parts of an already-mastered skill is a genuine productivity gain. Hints-only assistance appears to avoid the harm, while direct answers do not. Below are some references, from scientific studies to more accessible videos, if you want to dig into the topic.

• Kosmyna et al., Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task (MIT Media Lab, 2025): an EEG study showing reduced brain connectivity in LLM users, who also struggled to quote their own essays.
• Bastani et al., Generative AI without guardrails can harm learning: Evidence from high school mathematics (PNAS, 2025): students with an AI assistant performed better during practice, but 17% worse than controls on the subsequent unassisted exam; a "coach" mode giving hints instead of answers removed the harm.
• Ethan Mollick, Assigning AI: Seven Ways of Using AI in Class (One Useful Thing blog): practical patterns for using AI without outsourcing your thinking.
• JetBrains Academy, What AI does to the minds of novice coders (video): how AI can help or harm when learning to program, and why novices are the most at risk.
• Barbara Oakley, Don't Let ChatGPT Steal Your Brain (video): why real understanding still matters, and the most common mistake learners make when using ChatGPT.

ROS course

ECN Robotics Major

The goal of this course is to introduce the Robot Operating System. Labs are done on the Baxter robot, where students control some, then all of the degrees of freedom of the arms.

• Slides for ROS 2 are available here .
• Legacy slides for ROS 1 are still available here .
• The labs for ROS 2 are available on Github, download them to your ROS 2 workspace:
  git clone https://github.com/oKermorgant/ecn_ros2_labs
• Previous ROS 2 exams are available on Github. They usually boil down to writing a simple C++ node and some launch files.
  git clone https://github.com/oKermorgant/ros2_exams

Other ressources, given at Action Nationale de Formation, are available under the anf_launch package:

• Slides for numerous details on ROS 2 tooling (workspaces, colcon, environment variables, etc.)
• Slides for numerous details on ROS 2 launch files (syntax, difference with ROS 1, Python wrapper, etc.)

Task-Based Control

Master 2 CORO-IMARO

In this course the students are given state-of-the-art approaches for visual servoing and task-based control of robots.

The first lectures are about sensor modeling through interaction matrices and the study of the properties of the corresponding control laws (local minima, 3D trajectory, singularities).
First, the use of visual moments is detailed and compared to classical image-based and position-based approaches in visual servoing.
We then focus on redundancy and multi-task control, without then with unilateral constraints.

• The corresponding slides are available here .
• Additional slides from François Chaumette are on the GDR Robotique page
• Labs subjects are on Github:
  git clone https://github.com/oKermorgant/ecn_visualservo.git
  git clone https://github.com/oKermorgant/ecn_sensorbased.git
  git clone https://github.com/oKermorgant/ecn_baxter_vs.git

Modeling of Manipulators

Master 1 CORO

The goal of this course is to get knowledge on 3D geometry, serial robot modeling and basic control. We explore the Direct Geometric and Kinematic models, the Inverse Geometric Model, and joint-space and operational-space trajectory generation.

• The corresponding slides are available here .
• The book associated to the lecture is Modeling and Control of Manipulators, Part I, from Wisama Khalil.
• Interested readers can also get Part II, Dynamics and Control.
• Labs subjects can be downloaded from Github:
  git clone https://github.com/oKermorgant/ecn_manip.git

Vision for Robotics

ECN Robotics Major and Master 1 CORO

This course deals with computer vision with a focus on robotics applications. Students may come from the CORO program, or from the Robotics Major at École Centrale de Nantes.

• The book associated to the lecture is Computer Vision: Algorithms and Applications, from Richard Szeliski.
• Many of the slides come from the course of James Hays at Brown University: Introduction to Computer Vision.