Say Hi, Newton!
Welcome to our bi-weekly blog!
Our goal with this blog is to keep a log of how things are going for our Capstone project (every ~2 weeks), so in this introductory post, I’ll introduce the team members, what the goal of the project is & a summary of the phases for important milestones!
Let’s go!
The Team
We’re all undergraduate students studying at Concordia University, Montreal in a variety of majors.
| Member | Major |
|---|---|
| Shami Ivan Senga | Computer Engineering |
| Houssam Eddine Righi | Software Engineering |
| Augusto Mota Pinheiro | Computer Science (Biophysics) |
| Mohammad Umer | Mechanical Engineering |
| Camille Granade | Computer Engineering |
| Rayan Alkayal | Computer Engineering |
| Mohammed Bedair | Computer Engineering |
Goals & Objectives
We aim to develop a fully functioning quadruped robot and train it exclusively using various RL techniques. We intend to demonstrate the effectiveness of RL in robotic locomotion, while giving the bot some personality, and provide an educational platform for future students to add functionalities without needing extensive expertise in dynamics or control theory.
Project Phases
Phase 1
We’ve already started working on the project since a few weeks ago, with a starter project! We want to get a solid understanding of the pipeline for the project, from training to deployment on real robot, so we are repoducing a Capstone project from a team at the University of British Columbia (more information about it here) and we’re calling it: TWIP! In short, it’s a Two-Wheeled Inserve Pendulum robot (thereby its name) that is learning how to balance itself while idling & move around without tipping over.
Phase 2
The real meat of our project is in Newton: a 4 legged robotics educational platform for students & researchers to expand into whatever field they need. Building off the popular open source robot dog platform from the Open Dynamic Robot Initiative (ODRI), Solo12, we’re aiming to get a fully Deep RL (DRL) trained robot.
We have started analyzing the original Solo12 designs, refining them & adapting some concepts to our goals & objectives. The next steps consist in setting up the simulation environments, designing control systems, testing electrical components, etc.
Conclusion
So that’s it for an introduction! Make sure you follow this blog, because we’ll start posting some progress updates & learning experiences as we go along :)
Have fun!