Integrated framework for experiments with Mercator robots

The Mercator is a small mobile robot developed at IRIDIA on the basis of the Sphro RVR robot. It operates using ROS (Robot Operating System) to manage communication between its control software and the various sensors and actuators it is equipped with..

At IRIDIA, we develop control software for the e-puck platform using ARGoS: a state-of-the-art simulator for robot swarms that allows to use the same control software in both simulation and real robots experiments. In real wolrd experiments, we also use a ROS-based tracking system to monitor and evaluate the preformance of the robots.

The goal of this project is improve the integration of all these tools to enable reliable, easy to setup, and modular experiments with a swarm of Mercators. The basis of this work will be the existing infrastructure and ROS-based software.

Requirements: Good knowledge of C++ and Python and a lot of motivations to work with real hardware

Working language: English

Contact: Miquel Kegeleirs, and Mauro Birattari

Development and implementation of reinforcement learning on the AgileX Limo mobile robot platform for efficient navigation in simulated environments

The project centers on AgileX's Limo, a versatile multi-modal mobile robot, embedded with AI modules and ROS (Robot Operating System) packages. For the first time at IRIDIA, we want to explore the application of reinforcement learning on robotic platforms like the Limo. The primary goal is to train the Limo robot to navigate efficiently through simulated environments filled with obstacles, using reinforcement learning techniques.

The student will begin by training the Limo within a simulated environment created in Gazebo and will then transfer the acquired model to a real-world Limo robot. This project aims at gaining a thorough understanding and hands-on experience in reinforcement learning and its practical applications in robotics.

By the end of the project, the student will make a demonstration of the trained model's effectiveness in real-world applications with a real Limo robot.

Requirements: Good knowledge of C++ and Python and a lot of motivations to work with a robot simulator and with real hardware

Working language: English

Contact: Ilyes Gharbi and Mauro Birattari

Tracking and visualisation of a robot swarm with augmented reality

Tracking robotic systems is essential to assess their behaviour. At IRIDIA, we use an array of overhead cameras to locate the many robots that compose a robot swarm. We can then process the data off-line in order to extract some relevant metrics of the system, like its performance in a given task. However, real-time assessment of the swarm can also reveal interesting information, like the number of neighbours per robot, or the size of the largest robot cluster.

The goal of this project is to explore the use of augmented reality (AR) to track a robot swarm and visualise some of its features in real time. You will work with the VIVE Pro Eye virtual reality (VR) headset to both process stereo video from its front-facing cameras and render the desired visualisations. The use of a VR headset will allow users to focus on relevant parts of the environment and toggle whichever visualisations they consider relevant.

Requirements: Good knowledge of C++ and interest in working with VR/AR technologies

Working language: English

Contact: Guillermo Legarda Herranz and Mauro Birattari

Ceiling recognition for robot navigation and mapping

Navigation is an essential part of robotics. It gathers all the techniques a robot can use to explore the environment safely and efficiently. In particular, a robot should be able to gather information about its surroundings to take appropriate decisions about its direction. Nowadays, many robots use SLAM (simultaneous localization and mapping) to support the navigation, using information from LIDAR and cameras.

The Mercator is a small mobile robot developed at IRIDIA on the basis of the Sphro RVR robot. It is equipped with various sensors and actuators, including cameras. In particular, one of these cameras is pointing up and can provide informaion on the ceiling.

The goal of this project is to take advantage of the new point of view offered by this camera to improve the navigation capabilities of the Mercator. Using information from the ceiling, the robot could recognize the room it is currently exploring, as well as important landmarks such as doorframes.

Requirements: Good knowledge of C++ and Python and a lot of motivation to work with real hardware

Working language: English

Contact: Miquel Kegeleirs, and Mauro Birattari