Discipline: Computer Sciences and Information Management
Subcategory: Aerospace Engineering
Session: 1
Room: Private Dining
Oumar Toure - Fayetteville State University
Co-Author(s): Jonathan R. Soltren, Ahad A. Qureshi, David A. Riddy, Khali A. Crawford, Fayetteville State University, Fayetteville, NC.
A multi-agent system that allows a swarm of robots to carry out different tasks and work in conjunction with each other is ideal for the exploration of extraterrestrial terrain which is a goal inspired by NASA’s mission of robotic space exploration. Our team’s goal is to develop software and interface that will enable human controlled swarm robotics. A program that groups robots in a specific behavior model and gives each robot the functionality to detect objects is required. Robot Operating System (ROS) is our platform of choice. The ROBOTIS TurtleBot is a low cost, off-the-shelf robotic platform that utilizes the ROS framework. The Raspberry PI is a small single-board computer which is integrated with the Turtlebot to provide on-board computing ability. Raspberry Pi is a very versatile device that can accomplish a broad range of tasks that anyone can use to their benefit. Some of these tasks include browsing the internet, playing games, taking pictures, learning how to use a few programming languages (such as Python), and accomplishing a wide variety of projects. In this case, we plan to use the versatility of Raspberry PI to communicate and manipulate the Turtlebot for our benefit in the usage of human controlled swarm robotics.We hypothesize it is possible to create a swarm robotic AI (Artificial Intelligence) using a master node hierarchy, which delegates the tasks robots will take. The robots will concurrently execute code to reach a specified location. If the robot detects whether an object is in front of it, it should avoid the obstacle and continue its roam with the swarm. For object detection, the LIDAR (Light Distance and Ranging) will be used to create a map of the surrounding environment on the X-axis, it will not give an accurate account of the size and shape of objects within the Z-axis. DBSCAN and the K-Means algorithm will be used to anticipate the density of point distribution and create shapes from several clusters, respectively. The K-means algorithm identifies k number of centroids, and then allocates every data point to the nearest cluster, while keeping the centroids as small as possible. We believe that we can use both technologies simultaneously to create a more accurate map of the surrounding environment. For logic handling, a custom-made ROS node will handle all the conditions necessary to initialize a virtual map, connect each Turtlebot together as a cluster, and run distance-sensitive tasks to course correct each Turtlebot as they make their journey to a destination. The Turtlebots must send their position to the master hierarchy node, for a continuous autonomous movement. To avoid a Turtlebot from distancing itself from the group, a check will be run to recalibrate the Turtlebot cluster, keeping consistency in the swarm. In its current status, the code base can control the navigation of robots independently to execute complex navigation patterns. One of the primary issues that we are met with now is that of multirobot communication outside of simulation. There are many libraries containing packages for multirobot communication and mapping. However, most of these packages exist for older versions of ROS with some of these versions already meeting their End-of-Life support. There are a few solutions that could be implemented. One, the team could port one of the packages for the older versions of ROS so that they can be functional in the newer versions. A second solution would be to build the package from scratch, this would allow the team to directly tailor its functionality to our needs. The choice that fits our goals the best would be to build the package ourselves. The ongoing work is to develop visual object detection and navigation of the swarm around those obstacles. After reaching this technical objective, our project will enable technology for efficient and cost-effective terrain exploration with swarms.
Funder Acknowledgement(s): This work was supported in part by a NASA MUREP Space Technology Artemis Research (M-STAR) Implementation Grant. I also extend my thanks to Mark Paterson at NASA Johnson Space Center (JSC) for help.
Faculty Advisor: Sambit Bhattacharya, sbhattac@uncfsu.edu
Role: 1. Hardware engineering and integration. 2. Software programming for the robot.