Automatic path learning of an autonomous vehicle

    Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

    Abstract

    The automation level and flexibility of an autonomous test vehicle was increased by developing a teaching method that enables the vehicle to enter a partially unknown office environment, to learn its path automatically and to repeat it.The method was based on a coarse task description, consisting of simple motion primitives described by a motion control language developed for mobile robots.Natural landmarks were utilized as beacons to reduce the accumulative error in the estimates of position and heading maintained by dead reckoning navigation.The beacons were detected by ultrasonic sensors mounted on the vehicle.Distance travelled was measured by optical encoders and heading by a fibre optical gyroscope.Real-time obstacle avoidance was developed as basic ability to manoeuvre in a coarsely known environment.When the perceived obstacle was avoided, the vehicle was able to rejoin the original reference path.Obstacles were detected by a laser scanner based on a time-of-flight principle.The test results were encouraging and considered as a good base for further research and development in the field of autonomous navigation in a coarsely known environment.
    Original languageEnglish
    Title of host publicationSTeP 2000 : Millennium of Artificial Intelligence
    Subtitle of host publication'AI of today' : symposium on applications
    EditorsHeikki Hyötyniemi
    Place of PublicationEspoo
    PublisherFinnish Artificial Intelligence Society FAIS
    Volume2
    ISBN (Print)951-22-5128-0
    Publication statusPublished - 2000
    MoE publication typeA4 Article in a conference publication
    Event9th Finnish Artificial Intelligence Conference : Millennium of Artificial Intelligence - Espoo, Finland
    Duration: 28 Aug 200030 Aug 2000
    Conference number: 9

    Publication series

    SeriesPublications of the Finnish Artificial Intelligence Society. Conference Series
    Volume16
    ISSN1238-4658

    Conference

    Conference9th Finnish Artificial Intelligence Conference
    Abbreviated titleSTeP 2000
    Country/TerritoryFinland
    CityEspoo
    Period28/08/0030/08/00

    Fingerprint

    Dive into the research topics of 'Automatic path learning of an autonomous vehicle'. Together they form a unique fingerprint.

    Cite this