VSE knjižnice (vzajemna bibliografsko-kataložna baza podatkov COBIB.SI)
  • Samodejno zaznavanje ovir z RGB+D kamero za vodenje avtonomnega mobilnega robota [Elektronski vir]
    Hožič, David ...
    With increasing need of automation in the industry, many companies are utilizing autonomous mobile robots (AMRs). These need to be aware of obstacles in their planned motion trajectory. In this ... paper, we developed and evaluated obstacle segmentation and detection based on data from RGB+D camera attached to an AMR. The analysis pipeline consisted of camera calibration, FastSAM instance segmentation of the RGB data, enhanced via ByteTrack temporal filtering to reduce flickering, and coregistration of D or depth data, enhanced with filling operation to reduce gaps, so as to finally perform distance-toobject and object measurements calculations, and their inverse mapping to world coordinate space. Such information can be readily used for AMR navigation and was evaluated on videos acquired from an actual industrial AMR navigating in a production facility. Results showed high robustness of the proposed solution in detection of obstacles on the ground and above, and detection of humans. Albeit we did not perform any model re-training to the specific use cases, the proposed solution seems promising and ready for field testing.
    Vir: Zbornik triintridesete mednarodne Elektrotehniške in računalniške konference ERK 2024 [Elektronski vir] = Proceedings of the 33rd International Electrotechnical and Computer Science Conference ERK 2024 : Portorož, Slovenija, 26. - 27. september 2024 (Str. 567-572)
    Vrsta gradiva - prispevek na konferenci ; neleposlovje za odrasle
    Leto - 2024
    Jezik - slovenski
    COBISS.SI-ID - 209328899

    Povezava(-e):

    https://erk.fe.uni-lj.si/2024/papers/hozic(samodejno_zaznavanje).pdf

Vnos na polico