Specifications
AI Platform
NVIDIA Jetson AGX Thor
AI Performance (64GB)
Up to 1,200 TOPS
AI Performance (128GB)
Up to 2,070 TOPS
Memory Options
64GB or 128GB onboard
Camera Support
Up to 8x Orbbec Gemini 335Lg depth cameras, or up to 8x GMSL1/GMSL2 cameras
Ethernet
5x M12 10 Gigabit Ethernet with locking connectors
PTP Timing
Hardware IEEE 1588 precision time protocol support for sensor synchronisation
PoE
Power over Ethernet available on 4 of the 5 Ethernet ports
Industrial I/O
CAN, RS-232, RS-485, SPI-CAN and GPIO
Power Input
10 to 36 V DC
Protection Rating
IP65 sealed enclosure
Video
Advanced multi-stream encode and decode for simultaneous camera pipelines
Target Applications
Humanoid and service robots, AMRs, smart factory inspection
Integration Model
AI compute, high-speed networking, camera interfaces and industrial I/O consolidated on one unit
Overview
The TWOWIN T1530G is an edge AI computer built on NVIDIA Jetson AGX Thor, offered in 64GB and 128GB configurations rated at up to 1,200 and 2,070 TOPS respectively. It is aimed squarely at the physical AI category: humanoid and service robots, autonomous mobile robots, and intelligent machine-vision inspection on production lines.
What distinguishes it from a general-purpose Thor carrier is how completely the interfaces are integrated for a moving machine. The chassis carries five M12 10 Gigabit Ethernet ports — locking connectors, not RJ45, because a robot in motion will shake a latched plastic connector loose. Four of those ports supply PoE, so cameras and sensors are powered over the same cable, and the whole set supports PTP hardware timing, which is what lets multiple streams be fused into one consistent perception frame rather than approximate hand-alignment.
Camera support runs to eight sensors, either GMSL1/GMSL2 or Orbbec Gemini 335Lg depth cameras, with hardware multi-stream encode and decode handling the bandwidth. Alongside the networking sit CAN, RS-232, RS-485, SPI-CAN and GPIO, so motors, controllers and safety equipment connect directly instead of through adapters. Power is 10 to 36 V DC, and the enclosure is IP65.
The 128GB configuration is the one that matters for the emerging robotics software stack, where a vision-language-action model, a perception stack and a navigation stack may all be resident and running concurrently. Memory capacity rather than TOPS is usually the binding constraint in that scenario. QS Compute quotes both memory configurations alongside the Jetson AGX Thor module tiers and the GMSL2 camera ecosystem.
Key Benefits
Up to 2,070 TOPS from Jetson AGX Thor puts Blackwell-class compute on the robot rather than in a rack. 128GB configuration keeps VLA, perception and navigation models resident simultaneously. Five M12 10GbE ports with PTP and PoE give vibration-proof networking, sensor time sync and single-cable camera power. Eight GMSL2 or Orbbec cameras cover 360-degree perception. IP65 and 10-36 V DC input suit mobile platforms, outdoor sites and battery-powered machines. CAN, RS-485 and GPIO on board remove interface adapters from the design.
Applications
Humanoid and service robots; autonomous mobile robots in factories, warehouses and logistics centres; smart-factory defect inspection and product classification; production-line monitoring and intelligent machine vision; agricultural and outdoor autonomous machinery; research platforms for vision-language-action models.
Request a Quote — TWOWIN T1530G — IP65 JETSON AGX THOR ROBOT AI COMPUTER WITH 5X M12 10GBE
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →