UBTECH Releases Thinker-Sim, an Embodied Intelligence Simulation Platform
UBTECH has released Thinker-Sim, an embodied intelligence simulation platform centered on Real2Sim2Real, covering the digitization of real-world scenarios, large-scale synthetic simulation data generation, and robot capability evaluation. The platform improves rendering efficiency by 67.3%, can synthesize an average of 390 hours of simulation data per day on a single node (8 GPUs), and its self-developed communication module improves efficiency by more than 400%. The platform comes with built-in standardized task templates, and its visual workflow can adapt to more than 90% of target simulation scenarios with zero code.
Body · AI translation
Related topics top start Related topics top end Futures promotion begin Futures promotion end Miaopai begin Miaopai end Video player start Video player end Market chart begin Market chart end
Sina Tech News, November 2 morning news: Today, UBTECH's embodied intelligence simulation platform Thinker-Sim, centered on Real2Sim2Real, was officially released. It can build a one-stop technical system covering digitization of real scenarios, large-scale simulation data synthesis, and robot capability evaluation. It enables the real world to quickly enter simulation, massive data to be generated efficiently, and robot capabilities to be fully verified.
Addressing issues such as the high cost of real data collection, the complexity of building simulation environments, and the long algorithm verification cycle, Thinker-Sim provides three core functions: First, scene reconstruction and asset generation. Relying on the self-developed Thinker-GS high-fidelity scene reconstruction algorithm and the full-process SimReady asset automated generation technology, it realizes the conversion from real scene collection and 3D reconstruction to simulation asset generation, and supports scene reconstruction for industrial factories, commercial showrooms, and other scenarios. Second, simulation data synthesis. Relying on the self-developed high-performance rendering engine, large-scale parallel simulation architecture, and data synthesis algorithms, rendering efficiency is increased by 67.3%, a single node (8 GPUs) can synthesize an average of 390 hours of simulation data per day, and the self-developed high-performance communication module improves communication efficiency by more than 400%. Third, robot capability evaluation. It builds a unified simulation evaluation system for tasks such as perception, navigation, and manipulation, and reproduces the uncertainty of real environments through multi-dimensional randomization of scenes, lighting, textures, materials, and more.
In terms of product experience, the platform has built-in standardized simulation task templates, supports one-click startup and real-time screen streaming; through visual workflows, task orchestration can be built with zero code, adapting to more than 90% of target simulation scenarios; and based on the self-developed Ulab-Agent, it supports completing scene creation, environment configuration, task orchestration, and scheduling execution through natural language interaction.
UBTECH stated that Thinker-Sim will continue to consolidate the foundational capabilities related to embodied intelligence large models, providing infrastructure for the large-scale deployment of humanoid robots in industrial, commercial, and other scenarios. news_keyword_pub,stock,sz300024&sh603138 iframe below the main text iframe below the main text General statement General statement QR code at the end start QR code at the end start
QR code at the end start QR code at the end start Editor name and work code
Editor name and work code