Hardware#
This section provides information about the hardware requirements for Open AD Kit deployments, as well as the tested and planned hardware platforms.
Requirements#
Open AD Kit supports both amd64 and arm64 architectures. Requirements vary depending on whether you are running local development/simulation or deploying to a verified edge platform.
Local Development & Simulation#
For running deployments, simulations, and development workloads on a workstation or cloud instance:
| Resource | Minimum | Recommended |
|---|---|---|
| CPU | 8 cores | 16 cores |
| RAM | 16 GB | 32 GB |
| GPU | — | NVIDIA with 4 GB+ VRAM (for sensing/perception) |
| Storage | 50 GB | 100 GB+ SSD |
GPU Recommendation
An NVIDIA GPU is highly recommended for sensing and perception tasks. CUDA images require a working NVIDIA runtime; they do not automatically fall back to CPU. Use the standard CPU image or deployment configuration on hosts without a GPU.
Verified Edge Deployment#
For running a full Autoware stack on a verified edge platform, the requirements are higher:
| Resource | Specification |
|---|---|
| CPU | 40-core Arm Neoverse N1 equivalent (or better) |
| RAM | 32 GB |
| GPU | NVIDIA with CUDA support (hardware capability; see the ARM64 image note below) |
| Architecture | arm64 |
ARM64 sensing and perception
ARM64 edge hardware may include a CUDA-capable NVIDIA GPU, but the published
sensing-perception-cuda image currently supports linux/amd64 only. On
ARM64, use the standard sensing-perception image; sensing and perception
run on CPU and do not use the available GPU.
Why the difference?
The 40-core Neoverse N1 requirement reflects the verified ADLINK AADP-AVA platform, which runs the full Autoware stack with real-time constraints. Local development with demo simulations has lower requirements.
Tested Hardware#
| Platform | Architecture | Status | Notes |
|---|---|---|---|
| ADLINK AADP-AVA | arm64 (Ampere Altra, Neoverse N1) | Verified | Primary verified platform for edge deployment |
| ADLINK ADM-AL30 | arm64 | Verified | Used in Zenoh multi-vehicle fleet management demos |
| AWS EC2 G5.4XLarge | amd64 | Verified | GPU-enabled cloud instance for simulation workloads |
Tests Ongoing#
| Platform | Architecture | Status | Notes |
|---|---|---|---|
| NVIDIA Jetson Orin | arm64 | Tests Ongoing | JetPack 6 validation in progress. Not yet fully verified for production use. |
Development Hosts#
The following operating systems are supported for local development:
- Ubuntu 22.04 LTS (primary)
- Ubuntu 24.04 LTS
Other Linux distributions may work but are not actively tested.