Skip to main content

Getting Started with reComputer Robotics J601

The reComputer J601 is a compact yet powerful edge AI carrier board for Jetson AGX Thor, delivering up to 2070 TFLOPS. Built for development and production, it features M.2 Key E/M/B, 4x 10Gb RJ45, 4×USB 3.2, HDMI 2.1, 8×GMSL, and various IO's, ensuring seamless integration. It can be served as brain of humanoid. Supporting LLM & Physical AI frameworks like NVIDIA Isaac, Hugging Face, PyTorch, and ROS2/1, it bridges AI and robotics. With optimized real-time processing, it runs vision AI, transformers, and multimodal models, unlocking advanced AI for edge devices.

Features​

  • Supports NVIDIA Jetson AGX Thor T5000 and T4000 modules
  • Up to 2070 TFLOPS of AI performance
  • Up to 4x 10GbE RJ45 ports
  • 4x USB 3.2 Type-A ports at up to 10Gbps
  • HDMI 2.1 display output
  • M.2 Key M for PCIe Gen 4 NVMe 2280 SSD
  • M.2 Key E for M.2 2230 Wi-Fi modules
  • M.2 Key B for 4G/5G modules
  • Up to 8x GMSL2 cameras through two Mini-Fakra connectors and GMSL extension boards
  • Robotics I/O including isolated CAN, RS-232/422/485, I2C, I2S, GPI, and GPO
  • Wide-range 19V to 48V DC input through XT30
  • Software platform: JetPack 7.1

Specifications​

Module CompatibilityNVIDIA Jetson AGX Thor T5000 / T4000
PCB Size168 mm x 155 mm (without the Jetson AGX Thor module)
Display1x HDMI 2.1
USB4x USB 3.2 Type-A (10Gbps, Host), 1x USB 2.0 Type-C (Debug), 1x USB 3.0 Type-C (Recovery)
Ethernet4x RJ45 10GbE with T5000; 3x RJ45 10GbE with T4000
M.2 Key M1x M.2 Key M for PCIe Gen 4 NVMe 2280 SSD
M.2 Key E1x M.2 Key E for M.2 2230 Wi-Fi module
M.2 Key B1x M.2 Key B for 4G/5G module
Serial1x RS-232/422/485 (DB9 connector)
JST Ports4x CAN with T5000 or 2x CAN with T4000, 1x RS-485, 1x I2S, 1x I2C, 4x GPI, and 4x GPO
Audio1x microphone input and 1x audio output
Fan1x 4-pin fan connector (12V PWM)
RTC1x RTC 2-pin header
LEDs1x green PWR LED, 1x green SSD LED, and 1x RGB USR LED
Buttons1x Recovery button and 1x Reset button
GMSL2x Mini-Fakra connectors for up to 8x GMSL2 cameras
Operating Temperature-10°C to 60°C with thermal grease; -10°C to 55°C with a thermal pad
Power SupplyXT30, 19V to 48V DC
JetPackJetPack 7.1

Hardware Overview​

Side View 1
fig1
Side View 2
fig2
Bottom View
fig3

Flash JetPack​

Here, we will show you how to flash JetPack to an NVMe SSD connected to the reComputer Robotics J6014 / J6015. Both devices use the J601 carrier board, and the flashing procedure is the same.

Supported Module​

Prerequisites​

  • Ubuntu host PC
  • reComputer Robotics J6014 or J6015
  • NVMe M.2 2280 Internal SSD
  • USB Type-C data transmission cable
  • At least 220 GB of free storage on the host PC
info

We recommend using a physical Ubuntu host instead of a virtual machine. Seeed Jetson DevelopTool also supports Windows through WSL2, but a native Ubuntu host provides the most reliable flashing experience.

JetPack Version Ubuntu Version (Host Computer)
20.04 22.04 24.04
JetPack 7.1 ✅ ✅

Choose a Flashing Method​

Select either the graphical Seeed Jetson DevelopTool workflow or the command-line workflow below.

Seeed Jetson DevelopTool provides a guided graphical workflow that downloads, verifies, extracts, and flashes the firmware without requiring BSP commands. Install the tool by following the Seeed Jetson DevelopTool installation guide.

Video Tutorial​

Software Flashing Workflow​

Step-by-Step

Step 1. Launch Seeed Jetson DevelopTool and open Flash Center. Select reComputer J601 and JetPack 7.1 (L4T 38.4.0).

Step 2. Connect the host PC to the USB 3.0 Type-C flashing port. Press and hold the RECOVERY button, connect the 19V to 48V DC power supply through XT30, and then release the button after two seconds.

Step 3. Click Detect Device. Confirm that the connected Jetson module is detected (for example, AGX Thor T5000), and then click Next.

Step 4. Click Download / Prepare BSP. The tool downloads the firmware, verifies its SHA256 checksum, and extracts the BSP automatically.

Step 5. Click Start Flash and wait until the interface reports that flashing is complete. Do not disconnect the power supply or USB cable during this process.

caution

Flashing erases the data on the target NVMe SSD. Back up important data before you begin.

Step 6. Connect the reComputer Robotics J601 to an HDMI display and complete the initial system configuration.

info

Complete the System Configuration according to your needs after the first boot.

For detailed interface usage, please refer to Robotics J601 Hardware Interfaces Usage.

What Can You Do with J601?​

After you flash JetPack, explore the demo wikis below to see what you can build on reComputer Robotics J601. These cards are generated automatically from published Jetson Application and Other Devices wikis that mention J601 or Jetson Thor.

Fine-tune GR00T N1.7 for reBot Arm and Deploy on Jetson Thor
Physical AI

Fine-tune GR00T N1.7 for reBot Arm and Deploy on Jetson Thor

This wiki explains how to fine-tune NVIDIA Isaac GR00T N1.7 for the reBot Arm B601 DM and deploy it on Jetson Thor with TensorRT acceleration.

Fine-tune Isaac GR00T N1.5 for LeRobot SO-101 Arm and Deploy on Jetson Thor
Physical AI

Fine-tune Isaac GR00T N1.5 for LeRobot SO-101 Arm and Deploy on Jetson Thor

This wiki introduces how to get started with NVIDIA Jetson Thor and provides example workflows for deploying AI and robotics projects on Thor. Detailed step-by-step instructions and reference documentation are also provided.

Build a Four-Camera Fisheye Surround View Demo on Jetson AGX Thor
Computer Vision

Build a Four-Camera Fisheye Surround View Demo on Jetson AGX Thor

Build a four-camera fisheye surround-view demo on Jetson AGX Thor. Stitch a real-time BEV for chassis positioning, use YOLO to assist robot-arm grasping, and use a VLM to help understand the scene.

Deploy JoyAI-VL-Interaction on Jetson Thor
Generative AI

Deploy JoyAI-VL-Interaction on Jetson Thor

This wiki shows how to deploy JoyAI-VL-Interaction on NVIDIA Jetson AGX Thor, focusing on the minimal real-time video-language interaction stack for single-GPU edge deployment.

Control Rebot Arm with NemoClaw on Nvidia Jetson Thor
Physical AI

Control Rebot Arm with NemoClaw on Nvidia Jetson Thor

This article explains how to deploy NVIDIA’s NemoClaw/OpenClaw agent framework on Nvidia Jetson Thor and integrate the reBot Arm B601 robotic arm to build a local AI automation system combining visual perception, LLM reasoning, and robotic execution. It provides a complete setup guide covering hardware connections, NemoClaw and Ollama installation, robotic arm backend configuration, permission management, systemd auto-start services, and troubleshooting.

Voice Control reBot Arm B601 by Nvidia Jetson Thor
Physical AI

Voice Control reBot Arm B601 by Nvidia Jetson Thor

This wiki shows how to deploy a local voice-controlled robotic grasping system on NVIDIA Jetson Thor with the reBot Arm B601. It covers Jetson environment setup, Whisper speech recognition, Ollama local LLM deployment, OpenWebUI integration, GraspNet-based grasp planning, and voice-command control for robotic grasping tasks.

Control SO-Arm by OpenClaw on Jetson Thor
Physical AI

Control SO-Arm by OpenClaw on Jetson Thor

This wiki shows how to control SO-Arm on Jetson Thor with OpenClaw and LeRobot.

Resources​

Tech Support​

Thank you for choosing our products! We are here to provide you with different support to ensure that your experience with our products is as smooth as possible.

Loading Comments...