iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
To prepare an NVIDIA IGX Thor Developer Kit for robot AI, install IGX-SW 2.0 or later through the kit’s BMC-managed boot process, complete the on-device first-run setup, then install NVIDIA’s AI stack and explicitly assign containers to the intended GPU. The standard kit uses a monitor on the iGPU DisplayPort, a host computer connected to the BMC Ethernet port, and either a bootable USB installer or BMC Virtual Media.
This procedure applies to the standard IGX Thor Developer Kit. It does not apply to the Jetson AGX Thor Developer Kit, DRIVE AGX Thor Developer Kit, or IGX Thor Developer Kit Mini, which have different documentation and hardware procedures. NVIDIA listed IGX-SW 2.0.1, released October 7, 2026, as the latest production release identified here; check the IGX Download Center immediately before installation for a newer image.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
NVIDIA Jetson AGX Orin 64GB Developer Kit with Ethernet, USB, Display Port | $3,349.99 | Buy on Amazon |
| 2 |
|
NVIDIA Jetson AGX Xavier Developer Kit (32GB), 945-82972-0040-000 | $549.00 | Buy on Amazon |
| 3 |
|
NVIDIA Jetson Xavier Developer Kit (945-82972-0000-000) | $999.00 | Buy on Amazon |
What does an IGX Thor setup provide for a robot?
An installed IGX Thor gives you the platform operating system, real-time kernel option, boot firmware, and NVIDIA AI runtime needed to begin integrating a robot’s compute workload. It does not, by itself, configure cameras, lidar, motor controllers, robot middleware, inference models, timing budgets, or a functional-safety case. Those items depend on the robot design and the software supplied for its sensors and control system.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchNVIDIA positions IGX as an enterprise platform for industrial and medical edge applications. Its stated focus combines real-time sensor processing, AI reasoning, and functional safety for autonomous machines and robotics. Treat that positioning as the platform’s intended use; validate every safety and performance requirement on the complete robot.
#1 Best Overall
- The NVIDIA Jetson AGX Orin 64GB Developer Kit makes it easy to get started with Jetson Orin. Compact size, lots of connectors, and up to 275 TOPS of AI performance make this developer kit perfect for prototyping advanced AI-powered robots and other autonomous machines.
- The developer kit includes a Jetson AGX Orin 64GB module, and can emulate all the Jetson Orin modules. It supports multiple concurrent AI application pipelines with the NVIDIA Ampere GPU architecture, next-generation deep learning and vision accelerators, high-speed IO and fast memory bandwidth. Now you can develop solutions using your largest and most complex AI models to solve problems such as natural language understanding, 3D perception, and multi-sensor fusion.
- Jetson runs the NVIDIA AI software stack, and use-case specific application frameworks are available, including Isaac for robotics, DeepStream for vision AI, and Riva for conversational AI. You can save significant time with NVIDIA Omniverse Replicator for synthetic data generation (SDG), and by using NVIDIA TAO toolkit to fine-tune pretrained AI models from the NGC catalog.
- Jetson ecosystem partners offer additional AI and system software, developer tools, and custom software development. They can also help with cameras and other sensors, as well as carrier boards and design services for your product.
- With the computing capability of more than 8 Jetson AGX Xavier systems in a developer kit that integrates the latest NVIDIA GPU technology with the world’s most advanced deep learning software stack, you’ll have the flexibility to create tomorrow’s AI solution as well as today’s.
How do I confirm that I have the right kit and software?
Start by identifying the hardware and selecting an IGX software release that meets NVIDIA’s minimum requirement. The IGX Thor Developer Guide states that IGX-SW 2.0 or later is required for IGX Thor.
Identify the standard Developer Kit
- Use the standard NVIDIA IGX Thor Developer Kit instructions, not the Mini, Jetson AGX Thor, or DRIVE AGX Thor guides.
- Check the model and regional part number against the documentation and packaging. NVIDIA lists separate regional part numbers for IGX Thor Developer Kit variants.
- The standard kit overview identifies an NVIDIA IGX T7000 module, a 1,000 GB NVMe SSD, NVIDIA ConnectX-7 networking, a safety microcontroller, and a populated NVIDIA RTX 6000 PRO Blackwell MaxQ Workstation Edition discrete GPU. These are published configuration details, not independent benchmark measurements.
Select the current IGX-SW release
NVIDIA’s Download Center listed IGX-SW 2.0.1 with a release date of 2026-10-07. NVIDIA describes it as a minor update to IGX-SW 2.0 containing CVE and minor bug fixes, and directs users to install the IGX-SW 2.0 ISO followed by an APT upgrade. If the Download Center shows a later production release, use that release’s instructions and compatibility tables instead of assuming 2.0.1 is still current.
| Item | Requirement or published detail | Why it matters |
|---|---|---|
| Operating system | IGX-SW 2.0 or later; 2.0.1 was listed as released 2026-10-07 | Older images may lack required IGX Thor support or security fixes |
| Compute module | NVIDIA IGX T7000 module | Confirms that the standard IGX Thor software path is appropriate |
| Storage | 1,000 GB NVMe SSD | Provides the kit’s published local storage configuration |
| Graphics | Integrated GPU plus populated RTX 6000 PRO Blackwell MaxQ discrete GPU | Allows workload-specific GPU selection after setup |
| Networking and safety | ConnectX-7 networking and a safety MCU | Relevant to high-throughput sensor paths and safety-oriented system integration |
What should I gather before installation?
Prepare the physical connections, software files, and a recovery plan before powering on the kit. The standard installation uses the BMC for firmware inspection, power control, and optional remote media.
- A Windows, macOS, or Linux PC or laptop for creating installation media and opening the BMC WebUI.
- A monitor connected to the kit’s iGPU DisplayPort. Do not rely on the discrete-GPU display output for the initial installer screen unless NVIDIA’s documentation for your exact board says otherwise.
- A USB keyboard and mouse connected to the kit.
- An Ethernet cable from the host computer to the kit’s BMC port. NVIDIA’s guide identifies the BMC as the right-side port of the two Ethernet ports.
- The IGX OS ISO and the firmware bundle from NVIDIA’s IGX Download Center.
- A USB flash drive if you will use physical installation media, plus a suitable disk-imaging utility on the host computer.
- A stable power source and enough uninterrupted time for firmware updates and the operating-system installation.
Keep the firmware bundle available even if you expect to use automatic updates. Extract it when you need individual BMC, QSPI boot, or safety-MCU images for a manual update. Record the kit’s model, board revision, and current firmware inventory before changing anything.
How do I connect to the BMC for the first time?
The BMC provides out-of-band access for inventory, firmware, virtual media, and power operations. Use a direct Ethernet connection for initial setup so that unrelated DHCP or firewall behavior does not obscure the documented defaults.
- Connect the host computer’s Ethernet adapter to the kit’s BMC port, the right-hand port in the pair described by NVIDIA.
- Configure the host interface connected to the BMC with the static address
192.168.1.1. Leave other network interfaces disabled or disconnected temporarily if they create routing ambiguity. - Test reachability with a ping to
192.168.1.110. The guide documents that address as the BMC’s default IP; it also documents SSH access for troubleshooting. - Open
https://192.168.1.110in a browser and sign in to the BMC WebUI with the credentials supplied in the current NVIDIA guide. - After confirming access, replace default credentials according to your organization’s security policy. Treat factory credentials as temporary bootstrap access, not as an operating configuration.
If the address does not respond, verify that the host adapter really has 192.168.1.1, that the cable is in the BMC rather than the data-network port, and that the host firewall permits local ICMP or HTTPS testing. A successful link light alone does not prove the host and BMC are on the same IP subnet.
How should I inspect and update firmware?
Inspect versions before installing the operating system, then update only the components and image variant that match your board. The BMC WebUI exposes firmware inventory and the update workflow described in NVIDIA’s guide.
Recommended Free Tools
Review the inventory
- In the BMC WebUI, open the firmware inventory page and record versions for the BMC, QSPI boot firmware, and safety MCU.
- Compare each version with the firmware table for the IGX-SW release you selected. Do not compare against a table from a different Thor product or a different board revision.
- Determine whether your board uses ERoT or non-ERoT BMC firmware. The correct image is board-specific; never infer the file from its name alone.
Apply an update when required
- Download the firmware bundle associated with the selected IGX release and extract it if the WebUI requires an individual image.
- Choose the BMC update function in the WebUI, select the ERoT or non-ERoT image specified for your board, and wait for validation to complete.
- Start the update and keep the kit powered. Do not close the browser, remove the network cable, or interrupt power while the BMC is writing firmware.
- After the BMC reboots, reconnect to its address, recheck the inventory, and update any remaining components that the release table marks as required.
Firmware updates can temporarily remove BMC connectivity. If the page disappears, wait for the documented reboot interval before attempting another update. If inventory still reports an old version afterward, stop and verify the board variant and image type rather than repeating the flash operation blindly.
How do I install IGX OS from a USB drive?
A USB installer is the straightforward route when you are physically near the kit. It and the Virtual Media route use the same UEFI and GRUB installation screens.
- Use an imaging tool on the host computer to write the selected IGX OS ISO to a USB flash drive. Writing the image erases the drive, so confirm its device name before starting.
- Insert the completed USB drive into the IGX Thor kit and keep the monitor attached to the iGPU DisplayPort.
- In the BMC WebUI, open Operations → System power operations and reboot the Tegra.
- Watch the monitor. When the NVIDIA logo appears, press Esc to open the UEFI Boot Manager.
- Select the USB device that contains the installer. If more than one entry appears, choose the one corresponding to the inserted drive rather than a stale or unrelated boot option.
- At the GRUB menu, select Install IGX OS <version> with Real-Time Kernel.
- Allow the installer to run. NVIDIA says the installation takes about 20 minutes, depending on USB speed; do not treat that estimate as a guarantee.
- When the system reboots, let the QSPI boot-firmware capsule update finish. The guide says the capsule process runs for both A and B partitions. Interrupting it can leave the platform unable to boot.
How do I install through BMC Virtual Media instead?
Virtual Media is useful when the kit is remote or a physical USB port is inconvenient, but it depends on a reliable BMC connection and sufficient host-to-BMC throughput.
- Keep the ISO on the computer whose browser is connected to the BMC WebUI.
- Open Operations → Virtual Media and add the ISO file.
- Start the mount operation and wait for the WebUI to report that the virtual drive is attached.
- Use Operations → System power operations to reboot the Tegra.
- Press Esc at the NVIDIA logo, open UEFI Boot Manager, and select the OpenBMC Virtual Media entry.
- Choose Install IGX OS <version> with Real-Time Kernel in GRUB and allow the installation and QSPI capsule update to complete.
If Virtual Media is slow or disconnects, switch to a physical USB installer rather than repeatedly restarting the installer. A failed remote mount can leave the system at a boot menu without changing the installed OS, but an interrupted firmware capsule must be handled more cautiously.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteRank #2
- Newly updated version with an additional 16GB of memory for a total of 32GB of 256-bit wide LPDDR4X memory.
- NVIDIA Jetson Xavier is an AI computer for Autonomous Machines with the performance of a GPU workstation in under 30W
- The Jetson Xavier Developer Kit with Jetson Xavier module and reference carrier board is the fastest way to start prototyping with robots, drones and other autonomous machines
- Visit the NVIDIA Jetson developer site for the latest software, documentation, sample applications, and developer community information
- System Ram Type: Ddr Dram
What do I do during first boot and OOBE?
After the OS installation and firmware capsule process, NVIDIA’s on-device out-of-box experience configures the running system. Complete it on the monitor attached to the iGPU DisplayPort.
- Choose the requested locale, keyboard, time, and network settings.
- Create the local account and a strong password when prompted.
- Record the account details in your organization’s credential system; the BMC account and the operating-system account are separate.
- Finish OOBE and sign in to the installed IGX system.
- Confirm that the system clock, network route, and package repositories are correct before installing the AI stack.
At this point the kit is bootable and usable as an IGX computer, but no robot sensor driver, middleware graph, model, or safety configuration has been selected.
How do I install the NVIDIA AI stack and expose GPUs to containers?
Install the IGX AI package from the running system, then enable the CDI refresh service so container runtimes receive current device specifications.
sudo apt update
sudo apt install nvidia-igx-ai
sudo systemctl enable --now nvidia-cdi-refresh.service- Run
sudo apt updateto refresh package metadata from the configured repositories. - Install
nvidia-igx-ai. Resolve any repository or signing error before proceeding; do not substitute a package from a different Thor family. - Enable and start
nvidia-cdi-refresh.service. The service prepares the Container Device Interface specification used to request NVIDIA devices. - Restart or recreate test containers after the service has generated the current CDI data.
When launching a container, NVIDIA’s example selectors are --device=nvidia.com/gpu=0 and --device=nvidia.com/gpu=1. On the standard kit these indexes represent the iGPU and dGPU, respectively, in the documented example. Confirm the mapping on your installation and request the GPU that matches the application rather than assuming index zero is always the fastest device.
How do I adapt the installed platform to a robot AI workload?
Move from platform bring-up to workload integration in controlled layers: verify compute access, add sensor paths, connect robot middleware, deploy the model, and measure the complete system.
1. Validate the container and GPU path
- Run a minimal container that requests the intended CDI GPU device.
- Confirm that the container can enumerate the requested GPU and that CUDA, TensorRT, and framework versions are compatible with the installed IGX release.
- Check that the container sees the expected device memory and that no host process is consuming the resources needed by the inference service.
2. Add sensor capture
Install the camera, lidar, radar, or other sensor drivers specified by the robot vendor. Verify timestamps, frame identifiers, synchronization, packet loss, and transport bandwidth before adding inference. Sensor data that arrives late or with inconsistent clocks can invalidate an otherwise correct model.
3. Connect robot middleware and control outputs
Integrate the robot’s middleware, message definitions, and actuator interfaces according to their vendor documentation. Separate perception and planning containers from safety-critical control paths where the system architecture requires it. Define what happens when a sensor, model, network link, or GPU becomes unavailable.
4. Select the GPU deliberately
The standard kit includes both an iGPU and a discrete GPU. Choose between them using model memory demand, expected latency, concurrent workloads, sensor-I/O needs, and the available power and thermal envelope. This is an engineering decision framework, not a published comparative benchmark for the kit.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →5. Measure the real workload
Measure end-to-end sensor-to-decision latency, jitter, dropped frames, GPU utilization, memory headroom, thermal behavior, and recovery time under representative robot motion. Repeat measurements with the complete container set and network topology; a model-only test does not establish robot-level performance or safety.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which installation route should I choose: USB or Virtual Media?
Both routes reach the same UEFI and GRUB installer. Choose based on physical access and BMC reliability rather than expecting different operating-system results.
| Criterion | USB flash drive | BMC Virtual Media |
|---|---|---|
| Physical access | Requires someone to insert and remove media | Useful when the kit is remote |
| Network dependency | Only the BMC is needed for reboot and power control | Requires a stable host-to-BMC session while the ISO is mounted |
| Installer speed | Depends mainly on USB drive speed | Depends on the BMC and host network path |
| Failure recovery | Replace or rewrite the drive if its image is bad | Unmount and remount the ISO if the virtual session drops |
| Best fit | On-site installation or repeated local installs | Remote data-center or lab deployment |
What are the most common setup mistakes?
Using a different Thor product’s guide
Jetson AGX Thor and DRIVE AGX Thor are separate products with separate software stacks. A command or port description from one of those guides may not match IGX Thor. Verify the exact product name before downloading an image.
Rank #3
- 512-Core Volta GPU with Tensor Cores
- 8-Core ARM 64-Bit CPU
- 16 GB 256-Bit LPDDR4 memory
Applying Mini instructions to the standard kit
The Developer Kit Mini has different hardware and setup instructions. The procedure here assumes the standard kit’s BMC workflow, iGPU DisplayPort, and published hardware configuration.
Free tools Windows power users keep installed
One-click scans. No signup required.
Installing an old image
Check the IGX Download Center immediately before imaging the USB drive or mounting Virtual Media. IGX-SW 2.0.1 was the listed production release on 2026-10-07, but release status can change.
Choosing the wrong BMC firmware variant
Confirm whether the board is ERoT or non-ERoT and use the matching image from the release bundle. If the board identity is unclear, stop and obtain the correct hardware information instead of guessing.
Expecting containers to see a GPU automatically
Install nvidia-igx-ai, enable nvidia-cdi-refresh.service, and request the intended CDI device when starting the container. A container that starts without a device request may have no GPU access.
Interrupting capsule or firmware updates
Keep power and network connections stable during BMC updates and the post-install QSPI capsule process. If the system appears idle, allow the documented operation to finish before pressing reset.
How should I maintain and secure the kit after installation?
Change default BMC credentials, restrict BMC access to a management network, and keep a record of firmware and IGX-SW versions. Apply security and bug-fix updates according to NVIDIA’s release notes and your organization’s change-control process.
- Document the BMC IP, board variant, firmware inventory, OS release, kernel choice, and installed AI-stack package versions.
- Back up robot container manifests, model files, calibration data, and middleware configuration separately from the OS disk.
- Use signed, approved container images and review their GPU-device permissions.
- Monitor storage capacity, GPU memory, temperatures, and service restart behavior during long runs.
- Test recovery after a power loss and after loss of a sensor or network connection before operating the robot unattended.
A successful installation is only the starting point. Production acceptance requires evidence from the actual robot, sensors, model, control software, and safety architecture.
Frequently Asked Questions
Can I install IGX-SW 2.0.1 without a BMC connection?
The documented standard-kit workflow uses the BMC for firmware inventory, power operations, and either Virtual Media or reboot control. A physical USB can provide the installer, but you still need the BMC procedure described by NVIDIA for the normal bring-up and firmware checks.
Which display connector should show the installer?
Use a monitor on the kit’s iGPU DisplayPort for UEFI, GRUB, installation, and OOBE. The presence of a discrete GPU does not change that documented initial-display requirement.
How do I know whether a container is using the iGPU or dGPU?
Request the CDI device explicitly and inspect GPU visibility from inside the container. The standard-kit examples use device indexes 0 and 1; verify the mapping on your installation before assigning a production workload.
Does installing the NVIDIA AI stack install robot middleware too?
No. The AI package prepares NVIDIA compute and container-device integration. Sensor drivers, middleware, model files, calibration, actuator interfaces, and safety behavior must be supplied and validated for the specific robot.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
