swarm robots implementation: formation and collective behavior
Author
Pan, Yihan
Term
10. term
Publication year
2007
Pages
161
Abstract
Denne afhandling undersøger, hvordan små, perceptive sværmrobotter kan designes og programmeres til at opnå formation og kollektiv adfærd. Den præsenterer konstruktionen af nye mikrorobotter udstyret med flere IR-sender–modtager-par, omgivelseslyssensorer, en farvesensor og en global IR-modtager, drevet af et to-motorers drivsystem og koordineret af en mikrocontroller. Med udgangspunkt i disse hardwarefunktioner udvikles algoritmer til enkeltrobots navigation, herunder kommunikation, afstandsmåling, forhindringsundgåelse, tilnærmelse til lyskilder og farvebaserede reaktioner. Arbejdet udvides dernæst til samarbejde: formationskontrol for to robotter med forhindringsundgåelse designes, to tilgange til at holde formationen diskuteres og sammenlignes, og validering gennemføres både i simuleringer og fysiske eksperimenter. Endelig planlægges en kollektiv opgave og opdeles i to delopgaver (at finde og flytte “mad” samt at vende hjem med sikkerhedsvagt); rolle-specifikke algoritmer implementeres og demonstreres med to robotter. Simuleringer og indlejrede implementeringer anvendes gennemgående til at evaluere funktionalitet, og validering rapporteres for både simulerede og virkelige forsøg.
This thesis investigates how small, perceptive swarm robots can be designed and programmed to achieve formation and collective behavior. It presents the design and construction of novel micro-robots equipped with multiple infrared transmitter–receiver pairs, ambient light sensors, a color sensor, and a global IR receiver, driven by a dual-motor actuation and coordinated by a microcontroller. Building on these hardware capabilities, algorithms are developed for single-robot navigation, including communication, distance measurement, obstacle avoidance, light source approaching, and color-based reactions. The work then extends to cooperation: formation control for two robots with obstacle avoidance is designed, two approaches for maintaining formation are discussed and compared, and validation is carried out in both simulations and physical experiments. Finally, a collective task is planned and decomposed into two subtasks (finding and moving “food,” and returning home with a security guard); role-specific algorithms are implemented and demonstrated with two robots. Simulations and embedded implementations are used throughout to evaluate functionality, with validation reported for both simulated and real experiments.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents