Nav2CAN: Nav2 Context Aware Navigation
Translated title
Nav2CAN
Authors
Schmidt, Jonathan Eichild ; Schwörer, Tristan
Term
4. semester
Education
Publication year
2023
Submitted on
2023-06-02
Pages
77
Abstract
This thesis investigates how mobile robots can achieve context-aware navigation in human-populated environments and presents Nav2CAN: a modular system for ROS2/Nav2 that augments existing local and global costmaps with social and interaction proxemics, including support for F-formations when multiple people stand together. The system includes a local people-detection component but is designed to work with internal or external detection sources. The aim is to plan routes that respect social norms and minimize disruption, making robot behavior more predictable and comfortable for nearby people. Through simulations and module-level tests in real-world settings, Nav2CAN demonstrates that this approach enables human-aware navigation and can potentially increase human comfort around the navigating robot.
Denne afhandling undersøger, hvordan mobile robotter kan opnå kontekstbevidst navigation i menneskefyldte miljøer, og præsenterer Nav2CAN: et modulært system til ROS2/Nav2, der udvider eksisterende lokale og globale costmaps med sociale zoner og interaktionszoner baseret på proxemics, herunder F-formationer, når flere personer står sammen. Systemet omfatter en lokal persondetektionskomponent, men er designet til også at kunne anvende interne eller eksterne detektionskilder. Formålet er at planlægge ruter, som respekterer sociale normer og minimerer forstyrrelse, så robotter opleves mere forudsigelige og komfortable at omgås. Gennem simulationer og modultests i virkelige omgivelser viser Nav2CAN, at denne tilgang muliggør menneskebevidst navigation og potentielt kan øge menneskers komfort i nærheden af den navigerende robot.
[This apstract has been generated with the help of AI directly from the project full text]