INTRODUCTION

Mobile robots are designed for interacting with human environments and for working with humans on a daily basis.

To be considered of any use, these robots must exhibit some form of self-sustainability, so they must be capable of long-term autonomy.

In this work I develop a method for autonomously charging of a custom mobile robot that can be applied to several autonomous vehicles.

For the detection and tracking of the station, the robot employs a stereocamera (ZED) for the vision recognition of an Augmented Reality tag (AR tag) placed in correspondence of the docking station. In addition, the robot is equipped with two encoders sensors for extracting the actual velocities of the wheels. The two sensors data are fused for achieving a successful docking operation.

The NVIDIA Jetson Nano is the core of the robot. It executes the algorithm programs and sends the speed signals to the motor driver and the Arduino platform are responsible for driving the wheels. The robot is composed by two motorized wheels with TT DC motors and a caster one. The battery box and the power bank supply respectively the motors and the Jetson Nano. Moreover, it implements a GPS system that is part of the project of my colleague Andrea. 

As the software is regarded, the entire system is developed inside the ROS framework. The functions for the extraction and manipulation of the data, the detection and tracking of the AR tag are C++ nodes.

Now I'm going to explain some parts of this project: Simulation, Building the robot, Test the algorithm on the robot