Simultaneous localization and mapping (SLAM) is a key part of a robot’s toolbox for interacting with an external environment. The Walrus robot from the UT Austin Nuclear and Applied Robotics Group aims to be a heavy platform capable of self navigating through hostile environments while carrying heavy loads. It aims to keep users and personnel safe during its operation by being able to operate in environments that would naturally be unsafe or unsuitable for a person. My role in this project was to drive this robot around campus while also reporting on any discrepancies that occurred during the trial runs.

This robot consists of a metal cage attached to the top of a motorized Segway base. A lidar sensor acts as the robot's eyes and aids in perception of the environment. A long range radio transmitter allows the operator to interact with this robot from very far away. A water resistant and white container keeps the robot hardware cool and dry when operating in intense heat and rainy conditions.

The Walrus uses ROS as a framework for its system interactions. Critical software runs on the robot via ROS while acting as an interface for all other communications and connections. Internal SLAM software allows the robot to know its location while mapping its surroundings. Several trials were conducted in which the robot would be driven around UT campus while mapping the environment. At times, the robot would autonomously navigate to different waypoints provided by the user. Other times, it would follow a preset or recorded path from previous trials.

The Walrus was successfully able to map an environment while navigating through it. However, mapping errors occurred due to imperfect estimates of robot position. Additionally, the robot struggled at times to follow preset paths due to inherent odometry drift. Despite these failures, the robot was able to successfully complete all necessary tasks.

In conclusion, this research shows that a large and heavy robot vehicle can autonomously navigate through an environment with a accompanything payload. The ability of a human operator to use this robot remotely allows for increased safety when in dangerous or hostile environments. This experiment shows an ability of a robot to work in extreme and difficult environments.