Sun'iy intellekt va raqamli iqtisodiyotni rivojlantirish markazi
ICTWEEK 2026, Toshkent
Center for the Development of Artificial Intelligence and Digital Economy
ICTWEEK 2026, Tashkent
Bizning markaz O‘zbekistonda Jismoniy Sun'iy Intellekt (Physical AI) ekotizimini barpo etishni o‘z oldiga maqsad qilib qo‘ygan.
Our center aims to build a Physical Artificial Intelligence (Physical AI) ecosystem in Uzbekistan.
Sinfxonalar uchun mo‘ljallangan hamyonbop va ochiq kodli 6 o‘qli robot manipulyatori.
An affordable and open-source 6-axis robotic manipulator designed for classrooms.
Arzon robotlarni dars jarayoniga kiritish orqali qanday natijalarga erishamiz?
What results will we achieve by introducing affordable robots into the educational process?
SO-ARM101 manipulyatorini yig‘ish va teleoperatsiya jarayonlari bo‘yicha o‘zbek tilidagi qo‘llanma.
A practical guide in Uzbek on assembling and teleoperating the SO-ARM101 manipulator.
G‘oyadan jismoniy robotgacha, har bir bosqich keyingisini oziqlantiradigan uzluksiz sikl.
From idea to physical robot, an uninterrupted cycle where each stage feeds the next.
Har bir prototip haqiqiy robotga yuklanishidan oldin, NVIDIAning Isaac Sim va Isaac Lab muhitida minglab marta sinovdan o‘tadi.
Before deployment on a real robot, every prototype is tested thousands of times in NVIDIA's Isaac Sim and Isaac Lab environments.
Isaac Sim'da robot va muhitning fizika qonunlariga to‘liq mos raqamli nusxasi yaratiladi, har bir bo‘g‘im, sensor va yuzaga aniq ishlov beriladi.
A physically accurate digital replica of the robot and environment is created in Isaac Sim, meticulously processing every joint, sensor, and surface.
Yorug‘lik, tekstura va jismlar tasodifiy o‘zgartirilib, minglab virtual stsenariy generatsiya qilinadi, bu real dunyo xilma-xilligini taqlid qiladi.
Lighting, textures, and objects are randomly altered to generate thousands of virtual scenarios, mimicking real-world diversity.
GPU klasteri yordamida bir vaqtning o‘zida minglab robot nusxasi mustahkamlangan o‘qitishdan (Reinforcement Learning) o‘tadi, real vaqtdan ming barobar tezroq.
Using a GPU cluster, thousands of robot replicas undergo Reinforcement Learning simultaneously, thousands of times faster than real-time.
Simulyatsiyada shakllangan siyosat (policy) minimal moslashtirish bilan haqiqiy SO-ARM101 robotiga yuklanadi va sinovdan o‘tkaziladi.
The policy developed in simulation is loaded onto the real SO-ARM101 robot with minimal adaptation and tested.
Teleoperatsiyadan yig‘ilgan har bir harakat LeRobot ma'lumotlar to‘plamiga qo‘shiladi, VLA modelini o‘qitadi va yaxshilangan xatti-harakat sifatida robotga qaytadi.
Every action recorded from teleoperation is added to the LeRobot dataset, trains the VLA model, and returns to the robot as improved behavior.
Physical AI texnologiyalarini chuqur o‘zlashtirgan O‘zbekiston yoshlari ertaga haqiqiy aqlli robotlarni o‘z qo‘llari bilan yaratadi.
By deeply mastering Physical AI technologies, the youth of Uzbekistan will build truly intelligent robots with their own hands tomorrow.
Innovatsion ta'lim loyihalarida hamkorlik qilishga tayyormiz.
E'tiboringiz uchun rahmat!
We are ready to collaborate on innovative educational projects.
Thank you for your attention!