Undegraduate Research Project on Swarming Intelligence
Abstract
The goal of this project is to develop a swarm of intelligent agents using NXT Mindstorms robots to efficiently complete given tasks. This goal was achieved by developing a communication protocol and designing a sonar-based navigation system for multiple agents.
The communication protocol utilizes the Bluetooth API and object-oriented message routing for message deliveries to agents via host PC, serves as a communication hub, a network backbone for processing the robot's data and the decision-making logic. The swarm control system on the host PC is multithreaded, so it is scalable in terms of task parallelism. Low- and high-level communication protocol for multiple agents were developed and tested.
The navigation system is based on the ultrasound beacons approach that has several subsystems:
- 1) It utilizes an Arduino Mini microcontroller that was attached to the NXT robot and is responsible for receiving pulse data from the beacons and transmits data serially to the LEGO NXT Brick.
- 2) A transmitter interface was developed, which manages four ultrasound beacons with Piezo-electric transducers and an RF antenna that monitor the position of the robots.
- 3) A receiver interface was developed, which is attached to the robot and receives and processes the RF and ultrasound pulses.
- 4) Algorithms based on mathematical techniques were implemented to transform the robot's position based on the received data in the arena to the position coordinates in the XY plane.
Thus, we have developed an NXT Lego Mindstorms platform for studying intelligent swarming behavior.