SquadBot A Multi Agent Robotics Teaching and Research Platform

Multi-Agent robotics is the concept of multiple robots coordinating to complete one or more tasks. Such robotic systems are normally used when a single robot is not capable of accomplishing a task or group of tasks. Such systems have a lot of application in Field Robotics, both for civilian (agricul

Project Title

SquadBot A Multi Agent Robotics Teaching and Research Platform

Project Area of Specialization

Robotics

Project Summary

Multi-Agent robotics is the concept of multiple robots coordinating to complete one or more tasks. Such robotic systems are normally used when a single robot is not capable of accomplishing a task or group of tasks. Such systems have a lot of application in Field Robotics, both for civilian (agriculture, firefighting, etc.) and military (surveillance, reconnaissance, etc.) applications. We seek to cater to the problem of the availability of multi-agent robotic systems by introducing a multi-agent robotics platform for learning and research purposes. The aim is towards designing a platform that stands shoulder to shoulder with the existing multi-agent robotics systems, at a much lower cost with minimum reduction in capabilities. The primary focus in designing this platform is towards user-friendliness and scalability of the platform. Therefore, a high level of abstraction to the underlying hardware is maintained along with a modular design to ensure ease of usage and scalability. One key feature of the platform is that it is based on an open source meta-operating system, ROS, which promotes it as anyone can learn and operate the system without having to pay for learning the systems software dynamics. The technical topology of the project comprises of multiple robots along with a monitoring server, all of communication through a decentralized communication protocol. The reason for using the decentralized approach is to ensure that the platform is not completely dependent on the server, making sure that the system does not shut down completely in case there is a fault in the monitoring server. The robots in this platform have limited computation capacity and perform only simple tasks like data acquisition, communication and hardware actuation on board. All computationally intensive tasks like SLAM, Navigation and Path Planning are managed on the monitoring server. The currently implemented prototype is able to perform SLAM and Navigation on multiple robots, along with running autonomous exploration and multi-robot map merging algorithms on the platform. The project can be taken forward by constructing several more nodes for the system. These nodes should be able to execute standard xv multi-robot algorithms. In addition to this, implementation of a variety of multi-robot algorithms like exploration, convergence, etc. and bench-marking their performance against existing platforms can also be done.

Project Objectives

The project aims at building a low-cost, user-friendly and scalable platform for multi-robot research. The robot should have a modular and decentralized design, at an equal or lower cost than that of existing platforms while having capabilities equal or similar to the current expensive platforms. The user will be able to add as many nodes as they want to the system, limited only by the capacity of the WiFi network. Each node must be able to perform SLAM, odometry, be able to communicate data with other nodes and send relevant data to a stationary node. The project also aims at maintaining a high level of abstraction at the underlying layers to ensure that an inexperienced user can easily access the platform without extensive training. The three major objectives of the solution can, therefore, be concluded as follows:

• Low Cost: The overall cost of the project must be lower than the existing plat7 forms.
• User-Friendly: The system must have a simple user interface, little or no modifications should be required to operate the system.
• Scalable: The system must be able to support the addition of more nodes, without the limitation in software design.

Project Implementation Method

Keeping the objectives mentioned at the end of chapter 1 in consideration, the approach for building the system is highlighted below in figure 3.1. As already stated in chapter 2, a modular design is being used while maintaining a high level of abstraction at the user interface level. This helps to achieve a scalable and user-friendly platform where the user does not have to worry about the underlying hardware interfacing to use the platform. Furthermore, the communication system between the robots and the monitoring stations is designed to be decentralized. This allows for higher amount of fault tolerance in the case any particular node, especially the monitoring node shuts down.

Based on the requirements outlined in chapter 2 as well as considering the objectives mentioned in chapter 1, the following system overview has been developed, as shown in figure 3.2. The proposed solution is a three-layered architecture where the Blocks 1 and 2 run on the robot and Blocks 3 and 4 run on the monitoring server. The solution ensures that the system is easily scalable allowing for easy addition of more robots in the system. The block diagram will be discussed in detail in the next section.

As highlighted earlier, the system overview block diagram in figure 3.2 can be divided into three main subparts as follows:

• Blocks 1 and 2 in figure 3.2 which run on the robot, obtain data from the sensors and implement a PID controller on the motors based on the target velocities received. This combination of software and hardware is collectively called the base controller.
• Blocks 3 and 4 in figure 3.2 which run on the PC and run computationally intensive algorithms like SLAM, Navigation, Sensor visualization and any additional algorithm the user might want to run like map merging, exploration, etc. These blocks collectively form the monitoring server.

The decentralized communication protocol between the two parts stated above using ROS multimaster fkie package. Each of these three parts will be explained individually in detail in chapters 5, 6 and 7 respectively.

Benefits of the Project

Multi-Agent robotic systems have several applications in Field Robotics, both for civilian (agriculture, firefighting, etc.) and military (surveillance, reconnaissance, etc.) applications. The field of Multi-Agent robotics focuses on controlling many robots using either a centralized or a decentralized technique. It is currently an active research domain with regular contributions made to the domain from top research universities around the world. Unfortunately, there is little to no research being done in this domain in Pakistan now. Recently, HEC has funded the “Swarm Robotic Lab” which is one of the labs under the National Centre for Robotics and Automation (NCRA). NCRA was inaugurated in National University of Science and Technology on May 23rd, 2018. The lab aims at generating a generic scaling solution which will be capable of working for any application that fits the multi-agent robotic system criteria. Other than this mentioned lab, currently no lab is working in this domain in Pakistan right now.[1]
This can help in changing the technology sector in Pakistan and promote Robotics, Machine Learning, Artificial Intelligence and IoT due to the relation among these fields and our project being a platform which can allow the implementation of various projects/algorithms.

Bibliography:

1. http://web.uettaxila.edu.pk/uet/cped/SwarmRoboticsLab.asp

Technical Details of Final Deliverable

The block diagram for the base controller is shown in figure 5.1. In the block diagram the rectangles indicate computation platforms, ovals indicate software sub-processes and the hexagons indicate hardware interfaced to the computers.

Final Deliverable of the Project

Hardware System

Type of Industry

Education

Technologies

Robotics

Sustainable Development Goals

Quality Education

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com