Convergence of Multi-Robot Formation in ROS
Presenter Type
UNO Graduate Student (Doctoral)
Major/Field of Study
Computer Science
Other
Artificial Intelligence
Advisor Information
Hassan Farhat
Location
MBSC304 - G (Doctoral)
Presentation Type
Oral Presentation
Start Date
24-3-2023 10:30 AM
End Date
24-3-2023 11:45 AM
Abstract
This project is a continuation of the study in formation control in a distributed multi-agent system. To have some intelligent agents, such as a fleet of robots, to perform some jobs, sometimes we need them to come in formations. For a fleet of autonomous robots to form a pattern, a few pieces of information need to be reached agreement among robots: the number of robots in the fleet, the position of the desired formation, and the position of every robot in the fleet. Of course, we could specify which robot goes to the position we would want them to. Or we could provide GPS information for every robot in the system. However, this action will render the system into a centralized control scheme. A central controller would be an intuitive solution to control robots to form desired patterns. However, the job becomes more complicated when the number of robots in the fleet increases.
This work studies formation control in a distributed multi-agent system and addresses the problem where each robot could decide who goes where and how to get there. Our design lets the robots form a circle before forming the desired pattern. The circle formation settles their relative relationship into a stable status which is desirable for further formation configuration. A generalized control law has been developed in this work, which enables robots to form regular convex and some irregular convex patterns. And the current work is focused on a generalized solution for irregular convex and even concave formations. Some results will be present in simulations.
The transplant of the design into ROS is still under development. Some gaps still need to be mitigated between computer-based simulation and robotic-based implementation. Because we are developing a distributed solution, we could not use the GPS feature, which complicates the system design in many folds. By using the equipped distance detector on the robot and building a localized coordinates system, our latest design requires at least two reference objects. And recognizing and acknowledging a reference object itself is an active research topic.
Scheduling
10:45 a.m.-Noon, 1-2:15 p.m., 2:30 -3:45 p.m.
Convergence of Multi-Robot Formation in ROS
MBSC304 - G (Doctoral)
This project is a continuation of the study in formation control in a distributed multi-agent system. To have some intelligent agents, such as a fleet of robots, to perform some jobs, sometimes we need them to come in formations. For a fleet of autonomous robots to form a pattern, a few pieces of information need to be reached agreement among robots: the number of robots in the fleet, the position of the desired formation, and the position of every robot in the fleet. Of course, we could specify which robot goes to the position we would want them to. Or we could provide GPS information for every robot in the system. However, this action will render the system into a centralized control scheme. A central controller would be an intuitive solution to control robots to form desired patterns. However, the job becomes more complicated when the number of robots in the fleet increases.
This work studies formation control in a distributed multi-agent system and addresses the problem where each robot could decide who goes where and how to get there. Our design lets the robots form a circle before forming the desired pattern. The circle formation settles their relative relationship into a stable status which is desirable for further formation configuration. A generalized control law has been developed in this work, which enables robots to form regular convex and some irregular convex patterns. And the current work is focused on a generalized solution for irregular convex and even concave formations. Some results will be present in simulations.
The transplant of the design into ROS is still under development. Some gaps still need to be mitigated between computer-based simulation and robotic-based implementation. Because we are developing a distributed solution, we could not use the GPS feature, which complicates the system design in many folds. By using the equipped distance detector on the robot and building a localized coordinates system, our latest design requires at least two reference objects. And recognizing and acknowledging a reference object itself is an active research topic.