Is this project an undergraduate, graduate, or faculty project?
Undergraduate
Project Type
group
Campus
Daytona Beach
Authors' Class Standing
Marshall Yelvington, Sophomore
Lead Presenter's Name
Marshall Yelvington
Lead Presenter's College
DB College of Arts and Sciences
Faculty Mentor Name
Sergey Drakunov
Abstract
The Lunar Autonomy Challenge (LAC) is hosted by the John Hopkins Applied Physics Laboratory and NASA, and sponsored by Embodied AI and is for undergraduate students seeking to pursue controls theory and machine learning in the domain of robotics. We present a solution to enable our agent, or rover, to perform tasks such as traverse across a lunar environment, detect foreign objects, and localize. The agent is equipped with 2 stereo pairs, 4 additional stereo cameras, and an IMU. Using a 3D SLAM toolbox (RTAB-Map), an image processing package (NVIDIA Stereo Image Proc), and a landmark-based navigation controller (Nav2 and Dr. Drakunov’s algorithm), the agent can operate fully autonomously to create a geometric map of the area, and localize within it without the use of fiducials.
Did this research project receive funding support (Spark, SURF, Research Abroad, Student Internal Grants, Collaborative, Climbing, or Ignite Grants) from the Office of Undergraduate Research?
No
Lunar Autonomy Challenge: Enabling Autonomous Navigation in a Simulated Lunar Environment
The Lunar Autonomy Challenge (LAC) is hosted by the John Hopkins Applied Physics Laboratory and NASA, and sponsored by Embodied AI and is for undergraduate students seeking to pursue controls theory and machine learning in the domain of robotics. We present a solution to enable our agent, or rover, to perform tasks such as traverse across a lunar environment, detect foreign objects, and localize. The agent is equipped with 2 stereo pairs, 4 additional stereo cameras, and an IMU. Using a 3D SLAM toolbox (RTAB-Map), an image processing package (NVIDIA Stereo Image Proc), and a landmark-based navigation controller (Nav2 and Dr. Drakunov’s algorithm), the agent can operate fully autonomously to create a geometric map of the area, and localize within it without the use of fiducials.