Loomo Navigation

The main goal of this project was to make a Loomo robot capable of detecting and following a specific person. This was achieved using YoloR detector, DeepSORT algorithm as well as MediaPipe to detect a specific human pose. Before, conducting a robot race, the algorithm was developed to be launched on Google Colab using a laptop webcam.

Thymio Navigation

This project is implementing three algorithm to navigate a Thymio robot in a semi-static environment. The first algorithm is detecting the robot, fixed obstacles and goal using a webcam providing global view. The second is the A* path planning algorithm to planned the best path possible. Finally, the robot navigation is controlled with the help of an extended kalman filterto allow temporary webcam occlusion and movable obstacles avoidance using IR sensors.

E-Puck 2 Navigation

For this project the goal is to navigate an epuck 2 unicycle to a specific point. This point varying as it is defined by the source of emission of a 778 Hz sound. In addition, IR sensors are used to avoid obstacles on the vicinity of the robot while a TOF sensor is used to detect further obstacles and correct the trajectory. Note that there is no knowledge of the environment as the obstacles can have any shapes and position.