Autonomous Navigation for Quadruped Robots: Development and Optimization on the Unitree Go1 Platform

The growing adoption of service robotics in complex environments such as hospitals, industrial facilities, and public spaces highlights the need for versatile autonomous navigation. While some studies have addressed this challenge for specific robotic platforms, the application of such capabilities to different forms, like quadruped robots, remains relatively unexplored. This thesis, developed in collaboration with the LINKS Foundation, aims to integrate an autonomous navigation system for Unitree Robotics' Go1 quadruped platform, using ROS 2 (Robot Operating System) and its Navigation 2 stack. The project was structured in two main phases. The first phase focused on the general setup of the platform, with the aim of enabling communication between the different computers on board the robot and externally, allowing the remote execution of the software on external devices or in the cloud. Subsequently, the installation and configuration of the ROS 2 operating system was carried out on the robot, thus enabling the control of the robot itself and the readout from on-board sensors, including lidar, video camera and inertial sensors (IMU). In the second phase, the installation and configuration of the Navigation 2 stack was performed to enable autonomous robot navigation. Therefore, the system was implemented on the real robot to empirically evaluate its performance and autonomous navigation capabilities under real operating conditions. The results obtained confirm the achievement of the set objectives, laying the groundwork for future applications of quadruped robots in complex environments.