This thesis describes the prototyping and testing of an autonomous omnidirectional robot, called the OmniMaxbot. The OmniMaxbot is designed to carry cans of uranium ash at Cameco Corporation's Port Hope Conversion Facility while minimizing worker exposure. The robot makes use of the Robot Operating System (ROS) to allow the individual components to communicate as well as control the movement. In the course of this work, the OmniMaxbot's power distribution system was redesigned. The software for much of the hardware was developed, in part or in whole, in order to ensure the safe autonomous functioning of the system. Due to the excessive wheel slip caused by the OmniMaxbot's Mecanum wheels, laser scan matching was implemented to generate odometry data, as opposed to using encoder data. Different Simultaneous Localization and Mapping (SLAM) and autonomous navigation packages were tested, with the ROS hector_mapping, global_planner, and base_local_planner packages being selected for use. Configuration settings were determined that produced the best performance results. A variety of tests were performed to ensure that individual hardware, subsystems, and the full system worked as required. The results showed that the OmniMaxbot is capable of autonomously navigating to a pick-up location, picking up a mock ash can, navigating to a drop-off location, putting the can down, and returning to a standby location, all in an area with dynamic and static obstacles, without collisions.