AEC310
Atomrock Edge Computing AEC310
AEC310 is Atomrock's flagship edge computing platform, engineered for real-time perception and physical AI workloads at the source. Powered by NVIDIA Jetson Orin NX with up to 157 TOPS and built around a unified gateway architecture, it bridges on-site sensors, AI perception, and operational systems into one decision loop.
High-performance edge
Up to 157 TOPS
Multi-modal AI fusion
Industrial grade
High-performance Edge Computing
Powered by NVIDIA Jetson Orin NX (16 GB), featuring an octa-core CPU and a 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores. Together they deliver high-performance edge computing for real-time perception and physical AI applications, no cloud round-trip required.
Rich Front I/O
Multi-Modal AI Fusion
Atomrock's AI fusion capability correlates data from vision, audio, thermal imaging, and multiple sensors into a single operational view. Object detection, identification, tracking, and event-driven response all run from one aggregated stream of multi-source inputs in real time.
Multi-Sensor Partner Ecosystem
Atomrock collaborates with AI partners across hardware and sensing domains, including high-resolution cameras, EO/IR imaging systems, optical gas imaging (OGI) cameras, radar, and RF signal detection systems. The AEC310 ships with the integration APIs to bring all of them under one operational layer.
Gateway to Physical AI
A unified gateway that bridges AI intelligence with physical-world execution at the edge. AEC310 connects on-site sensors, AI perception, and operational systems to convert real-time data into actions, enabling situational awareness, decision-making, and response in complex physical environments.
Rugged, Certified Hardware
Industrial-grade aluminum enclosure with fanless passive heat dissipation, rated for 24/7 operation across -25°C to 70°C. MIL-STD-810H certified for shock and vibration; CE / FCC / EN50155 / EN50121-3-2 compliant for railway and industrial environments. Wall-mountable with optional DIN rail.
Powered by AtomOS 4.0
The AEC310 edge AI platform is powered by AtomOS 4.0, a purpose-built operating system designed with security as a foundational principle rather than an add-on layer. The OS employs a deliberately stripped-down architecture with rigorously screened and whitelisted programs, locked-down ports and protocols, and a strict policy of storing no security-sensitive data locally on the device. All external communications are secured through encrypted connections, while the entire system is hardened from the OS layer upward, creating a significantly reduced attack surface suitable for enterprise, critical infrastructure, and regulated industry deployments.
Secure Provisioning at Scale
Beyond its security posture, AtomOS 4.0 delivers operational advantages through native self-provisioning capabilities that enable secure, automated onboarding and configuration at scale. This allows large numbers of distributed edge devices to be deployed rapidly with minimal manual intervention while maintaining consistent security policies across the fleet. The combination of OS-level hardening, encrypted communications, and streamlined provisioning makes the AEC310 particularly well-suited for organizations requiring both high-assurance edge intelligence and manageable, policy-driven deployment across multiple sites.
Application Scenarios
Nine domains powered by 157 TOPS of on-device AI, from defense and infrastructure to manufacturing and disaster response.
Aerial Threats Defense
Counter-drone perception, classification, and tracking for protected airspaces.
Maritime & Coastal Defense
Wide-area perception across coastal infrastructure, ports, and shoreline security.
Smart City Safety
Citywide camera network analytics for crime prevention, incident response, and public safety.
Infrastructure Safety
Bridges, tunnels, power stations, and other critical infrastructure with continuous monitoring.
Office Management
Access control, occupancy analysis, and incident detection across enterprise facilities.
Energy Management
Substation perimeter security and equipment health monitoring for grid and energy assets.
Manufacturing Defect Inspection
Real-time visual inspection on production lines, catching defects the eye misses.
Gas Leak Detection
Optical gas imaging integration for refineries, chemical plants, and hazardous sites.
Thermal Imaging Disaster Response
Survivor heat-signature detection and hazard visualization for emergency operations.
Specifications
| SYSTEM | |
|---|---|
| CPU | 8-core Arm Cortex-A78AE v8.2 64-bit CPU |
| GPU | 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores |
| System Support | AtomOS 3.9.0+, VisionSense AI package |
| I/O INTERFACE | |
|---|---|
| USB | 1 USB 3.1 Gen 2 · 2 USB 3.1 |
| Serial | 2 COM RS-232 / 422 / 485 |
| GPIO | 7-bit GPIO |
| Micro USB | 1 Console Port · 1 Recovery Port |
| Button | 1 Power · 1 Force Recovery · 1 Reset |
| LED | Power, SSD |
| SIM | 2 SIM Card Socket |
| Antenna | 6 Antenna for WiFi / 4G / 5G / LTE / GPRS / UMTS |
| EXPANSION | |
|---|---|
| M.2 | 2 M.2 Key B Socket (3042 / 3052, USB3) · 1 M.2 Key E Socket (2230, PCIe / USB) |
| GRAPHICS | |
|---|---|
| Interface | 1 Digital Display, up to 3840 × 2160 @ 60Hz |
| Video Encode | HEVC: up to 1× 4K@60, 3× 4K@30, 6× 1080p@60, 12× 1080p@30 · H.264: 1× 4K@60, 2× 4K@30, 5× 1080p@60, 11× 1080p@30 |
| Video Decode | HEVC: up to 1× 8K@30, 2× 4K@60, 4× 4K@30, 9× 1080p@60 · H.264: up to 1× 4K@60, 2× 4K@30, 5× 1080p@60, 11× 1080p@30 |
| ETHERNET | |
|---|---|
| LAN1, LAN2 | 10/100/1000 Base-T Ethernet GigE LAN, RJ45 connector (optional X-coded M12) |
| POWER | |
|---|---|
| Power Input | DC 9V to 50V |
| Power Interface | 3-pin Terminal Block (V+, V-, Frame Ground) |
| Ignition Control | 16-mode Software Ignition Control |
| Remote Switch | 3-pin Remote Switch Terminal Block |
| STORAGE | |
|---|---|
| M.2 | 1 M.2 Key M 2280, up to 1TB NVMe SSD pre-installed |
| SD | 1 Micro SD (External) |
| MECHANICAL | |
|---|---|
| Dimensions | 193 × 134 × 55 mm (7.59" × 5.28" × 2.17") |
| Weight | 1.7 kg (3.74 lb) |
| Mounting | Wallmount · DIN Rail (optional) |
| ENVIRONMENT | |
|---|---|
| Operating Temp. | 15W TDP: -25°C to 70°C (with 0.63 m/s air flow) · 25W TDP: -25°C to 55°C (with 0.63 m/s air flow) |
| Storage Temp. | -40°C to 85°C |
| Humidity | 5% to 95%, non-condensing |
| Relative Humidity | 95% @ 70°C |
| Shock | Operating, MIL-STD-810H, Method 516.8, Procedure I |
| Vibration | Operating, MIL-STD-810H, Method 514.8, Procedure I, Category 4 |
| EMC | CE · FCC · EN50155 · EN50121-3-2 |
| AI FUNCTIONS | |
|---|---|
| Advanced AI | Video AI (VisionSense) · Audio AI (AudioSense) |
| Physical AI | Object detection, identification, and tracking across physical environments |
| Customized AI | Scenario-specific AI customization based on application environments |
| Atomrock Integration API | APIs for seamless integration with Atomrock's ecosystem partners and external systems |
| CLOUD / VIDEO / AUDIO | |
|---|---|
| Cloud Connection | AEC cloud connectivity |
| Cloud Recording | Cloud recording with 7 to 366 days data retention |
| AI Events | Cloud retention of AI event logs and snapshots |
| OPTIONAL | |
|---|---|
| TPM | SLB 9673 XU2.0 FW26.10 with I2C interface |