Sanoat robot manipulyatorlari uchun chuqur o‘qitishga asoslangan ob’yektlarni aniqlash va ushlash tizimini ishlab chiqish

2026

Closed access Metadata (CSL-JSON)
DOI 10.65322/j.srd.03172026.25
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Ushbu maqolada sanoat robot manipulyatorlari uchun chuqur o‘qitish algoritmlariga asoslangan tasvirli boshqaruv tizimi taklif etilgan. Tizim ikki bosqichli neyron tarmoq arxitekturasidan iborat bo‘lib, obyektlarni real vaqt rejimida aniqlash uchun YOLOv8 modeli va ushlash pozasini baholash uchun GG-CNN modeli qo‘llanilgan. Taklif etilgan tizim PyBullet simulyatorida 6 erkinlik darajali UR5 robot manipulyatori bilan integratsiyalashtirildi va 5000 ta eksperiment orqali sinovdan o‘tkazildi. Olingan natijalar obyektlarni aniqlashda 94,2% mAP@0,5 va ushlash muvaffaqiyat darajasi 87,3% ga erishilganini ko‘rsatdi. Tizimning umumiy inference vaqti 42 ms ni tashkil etadi, bu real vaqt rejimida ishlash uchun yetarli. Sim-to-real transfer muammosini hal qilish maqsadida domain randomization usuli qo‘llanildi va u real sharoitdagi samaradorlikni 39,2% dan 84,5% gacha oshirishga imkon berdi.