Exploration of a Custom Low-Cost Autonomous Unmanned Ground Vehicle Using State of the Art Artificial Intelligence Techniques
We present the development of a UGV that is modified to be a "Mapping Master "of a mapping unit system. The Mapping Vehicle will provide computing power and the localization as reference for the smaller agents. This mapping vehicle is built to be low-cost, compatible with ROS and have sufficient payload to carry and implement various sensors. The main objective of the robot is to navigate through outdoor and indoor environments with precision and speed using sensory feedback. With the implementation of the components, the paper will delve deeper into the three important requirements that the UGV will implement to navigate in an unknown environment:1. Mapping System: which senses and understands the environment the system is in2. Localization System: provides a frame of reference in Cartesian coordinates for the robot's current location3. Obstacle Avoidance System: that keeps the vehicle from running into obstacles, other vehicles, and keeping the platform in a bounded region. Due to the limitations of a physical system, such as only having so much power to run in tests, errors in control dynamics, and constantly changing environments, we emulate the system in NVIDIA Isaac Sim for physical simulations such as running tests for Reinforcement Learning.