Robust Potential Field-Based Communication & Control Architecture for Robotic Swarms - Using the nRF52 System-on-Chip and a mesh network topology
Master thesis
Permanent lenke
http://hdl.handle.net/11250/2559482Utgivelsesdato
2018Metadata
Vis full innførselSamlinger
Sammendrag
Which technologies and algorithmic design are suitable for both the control strategyand the communication protocol to ensure robustness within a robotic swarm?
The aim of this study is to answer this question. The study is performed by obtaining anunderstanding of the field of swarm robotics and proposing a robotic swarm system withrobustness in mind. The robotic swarm system contains possible solutions for the controlstrategy, the communication protocol, hardware design and an in-room localizationmethod. The innovative part of this swarm system is the idea of joining communicationand control in a cooperative manner, based on previous studies on different robotic swarms.The communication protocol was designed to establish a mesh topology using the Threadprotocol, MQTT-SN and ROMANOs. ROMANOs is a new application overlay protocolbased on ROMANO, and is first introduced in this project. The control strategy introducesa potential-field based PID-controller, designed for its efficiency and practical approach.The cooperation between communication and control is the fact that the estimation ofsignal strength values will directly affect the potential fields of the control strategy. Thecontrol strategy will then initiate control actions to maintain the communication network.The hardware design platform consists of two printed circuit boards that houses all necessaryelectronics to realize the swarm algorithm, and a 3D-printed cylindrical chassis. Thein-room localization method is based on ranging measurements from laser sensors, signalstrength estimates and heading computations based on magnetometer data. A listeningdevice consisting of a Node.JS MQTT module is created in order to acquire performancedata from the swarm which is stored in an SQL database and further analysed in MATLAB.
The proposed swarm application yields a mostly favorable, but mixed result. The jointcontrol- and communication algorithm is successful in finding a local minima within thenetwork and maintaining said network. However, some parts of the algorithm are taxingand could be further refined. The communication protocol is only able to maintain amessaging frequency of up to 5 Hz for a six-member swarm, and the processor is only ableto run the proposed swarm algorithm at a maximum of 50 Hz. What measures that shouldbe taken to mitigate this and future work for this swarm application will be proposed.Nevertheless could the proposed swarm application work as an important stepping-stonefor a true robust swarm application in the future.
Keywords : robotic swarms, potential-field based control, thread mesh network, MQTTSN,ROMANO, laser distance measurements, magnetometer, joint control- and communicationalgorithm, PCB design, algorithm design, robust swarm application