Autonomous mobile robots in agriculture have brought digital transformations with new automation technologies. Mundane tasks in agricultural fields involve long-term deployment of scheduled assignments done over long working hours. Hectares of land also require much distance to be traveled by the robot, wherein surviving without stranding powerless amid operation is a key factor towards long-term autonomy. Scheduling tasks are usually required as the mobile robot’s battery life is limited. However, dynamic and unstructured outdoor environments make it practically challenging to abide by a schedule even with fixed charging routines. Most robot systems use hard thresholds and fixed periods to decide when to charge. These measures produce behaviors that ignore the safety risk by not considering current conditions. Instead, the robot should adaptively return to the charger or call for human assistance. Safety was a score in this work that represented the robot’s ability to return to the charger at any time. This thesis proposes a real-time safety watchdog that exploits parameters like charger distance, mileage, and current battery status to calculate safety. Hence, a safety-constraint based approach is followed to flexibly sustain survivability during long-term deployments.This work compared a typical rule-based system to the real-time safety watchdog that monitors multiple parameters based on the battery state. Simulated results empirically showed that the latter approach ensures survivability and maintains a healthy performance that supports long-term autonomy. The concept of a real-time safety watchdog is also showcased on a physical robot, pushing statistical evaluation that uses values of long-term deployment as future work. Although a watchdog ensures survivability to a planned schedule, adapting the experience in the upcoming schedule can be worked upon to continue this work. ; Autonoma mobila robotar inom jordbruket har lett till digitala förändringar med ny automationsteknik. Vanliga arbetsuppgifter inom ...


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Towards Long-Term Autonomous Survivability in Agriculture : A real-time safety watchdog to ensure the survivability of an outdoor plant monitoring robot. ; Mot långsiktig autonom överlevnadsförmåga inom jordbruket : En säkerhets watchdog i realtid för att säkerställa överlevnaden hos en robot för övervakning av växter utomhus.


    Beteiligte:

    Erscheinungsdatum :

    2022-01-01


    Medientyp :

    Hochschulschrift


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629



    Watchdog controller

    JEREMY GREENWOOD / ROBIN BOYD / CHRIS CLARKE | Europäisches Patentamt | 2018

    Freier Zugriff

    WATCHDOG-STEUERGERÄT

    GREENWOOD JEREMY / BOYD ROBIN / CLARKE CHRIS | Europäisches Patentamt | 2019

    Freier Zugriff

    Watchdog for monitoring a processor

    SUDHAUS ANDRE / SUBIJANTO TAN | Europäisches Patentamt | 2020

    Freier Zugriff