Axel Paris

The project is built as a series of milestones, each one a live demo on the robot. We only move on once the previous one works: locate still objects on the conveyor, track them and predict their position, move the gripper to a still object, then follow, intercept and grab a moving one, and finally sort it by colour.

Constraints: the system has to be reliable and repeatable, and stay within the robot's joint limits and motor heating. Success rate will be measured over test series and compared with the state of the art.

Results to come.

A ROS 2 / Gazebo simulation of our competition robot, used to test localisation and trajectory following before running anything on the real hardware (Raspberry Pi).

The point was to unblock the software team: with a simulated robot, we could develop everything downstream of localisation (trajectory following, game strategy…) without waiting for the mechanical and electronics teams to deliver the real robot.

A Franka Panda arm uses a depth camera to find coloured cubes on a table. It picks them up and stacks them in a chosen order. Before grasping, the gripper rotates to match each cube's angle. If a cube is blocked by another one, the arm moves that cube out of the way first.

I wrote the inverse kinematics (damped least squares on the full pose). The motion is tracked by a model predictive controller (MPC) running at 100 Hz.

Mechanical design and control of a robotic hand.

Five fingers actuated by fishing-line tendons, each pulled by a Miuzei servo. An ESP32 drives the servos, and the commands come from a Python program using OpenCV.

A small web game: you have to find a hidden city. Each guess tells you which direction the target lies from the city you entered, so you close in on it step by step.

I'd played a game like this online, couldn't find it again, so I rebuilt it myself.

A two-wheeled robot that stays upright on its own. Most of the work went into the control loop.

Sensor (IMU), controller (PID), loop rate, tuning.

An arm designed and built from scratch: mechanical parts, simulation, then control through ROS 2.

Actuators, sim-to-real link

My first hands-on build: a small 3D-printed car with its own electronics and code. The goal was simply to learn, by wiring, programming and printing everything myself. It's where most of my later projects started.

A web app that turns 2D photographs into 3D models for surgical use.