Alex Robotics

Learning material

          
            Commande optimale et apprentissage par renforcement (Notes de cours) 
        
           Commande
            optimale et apprentissage par renforcement (Vidéos)
          
           Pyro: a python toolbox for robotics 
        
           Robotique: cinématique, dynamique et commande
            (Vidéos) 
        
            Robotique: cinématique, dynamique et commande (Notes de cours)
        
            Tutoriels interactifs (Colab)
        
           Systèmes asservis (Vidéos)

Colab Tutorials

Optimal control:

Optimal control overview
This page introduce various optimal control algorithms for solving a pendulum swing-up task.
LQR vs. Dynamic Programming
This page shows a quick demo of how DP (dynamic programming) can be used for finding global optimal control policy of non-linear systems.
Trajectory optimization
This page shows and exemple of planning an optimal trajectory for a cart-pole swing-up with the direct-collocation method.
Optimal swing-up with a custom cost function
This page allow for quickly modifying the cost function to see how it influences the solution.
Trajectory optimization with Casadi
This page is an interactive tutorial using casadi for trajectory optimization
Minimum time optimal policy for a mass
This page is an interactive tutorial demonstrating using the value iteration algorithm.
Policy evaluation for a minimal time problem
This page is an interactive tutorial demonstrating using a numerical algorithm to evaluate a custom policy
LQR for finite vs infinite horizon
This page is an interactive tutorial demonstrating 2 variations of a LQR controller for the cartpole system

Robot arm control:

2 DoF robot controlled with a PID
This page allow for playing with the PID gains and animating the results.
2 DoF robot controlled with a custom controller
This page is a template for implementing and testing controllers for a two-link planar manipulator.
2 DoF robot arm kinematic controllers
This page introduce the concept of controlling the end-effector of a robot when its joint are controlled with velocity set-points.
2 DoF robot arm impedance controllers
This page introduce the concept of impedance control in both the joint-space (angles) and the task-space (x,y,z).
5 DoF robot arm kinematic controllers
This page introduce to the control of redundant manipulators

Classical controls:

PID on a 2nd order system
This page is a template for testing a PID controller on a two-mass-spring-damper system.
Phase Plane Analysis
This page is an interactive tutorial on phase plane analysis for dynamical systems.

Machine Learning:

Diffusion in 2D
This page is an interactive tutorial on the base concept of diffusion.
Diffusion policy
This page is an interactive tutorial on the base concept of diffusion policies.

Dynamic:

Planar Plane Dynamics
A colab page introducing the planar plane class in pyro.

Reinforcement learning:

Drone learning to fly with PPO
A colab page demonstrating learning a flight controller with the PPO algorithm.
Cost-to-go approximation
A colab page demonstrating various function approximation for the optimal cost-to-go of a pendulum swing-up problem

Pyro toolbox:

Pyro Tutorial no1
The Dynamic System class and basic functionnality
Pyro Tutorial no2
Creating a custom dynamic class
Pyro Tutorial no3
Closed-loop systems and the controller class
Pyro Tutorial no4
The manipulator robot class

UdeS Classes

Robot kinematic : GRO203
Linear algebra : GRO211
Controls : GRO501
Robot differential kinematic and quasi-static controls: GRO640
Robotic : Modeling and Controls - GMC714
Optimal control and reinforcement learning: GRO860