Xplorer Series – Rugged Industrial Edge IoT/AIoT Controller & Embedded Mission Computers – Raspberry PI CM4/CM5, Google Edge & Hailo AI Ecosystems
www.austral-elec.com - Intelligent Technologies for Marine, Industrial IoT and Unmanned Vehicles
Xplorer4 & 5 Features – Wall mounted CNC black anodized aluminium enclosure
Ready for the Compute module CM5, Matter, Starlink Direct To Cell, New Generative AI modules…
- SWaP-C Optimization: Cost-effective, industrial, rugged, waterproof and dustproof IP67, wide temperature range, miniaturized, very lightweight, very low power consumption, fanless…
- Industrial, Marine & Vehicle field connectivity:
– M12 Connectors with cost-effective on the selves cables
– Industrial Gigabit LAN Ready for IP Camera, Modbus TCP, Profinet, ETherNet/IP, BACnet, OPC-UA, MQTT, DDS, Websocket, ZMQ, UDP, TCP, NMEA OneNet…
– Multiples CANbus FD, RS232, RS485, Modbus, some with galvanic isolation, Digital and Analog I/O.
– And more connectivity using M.2 expansion. - 🏆 Best in class wireless connectivity capability: Wifi, Bluetooth, GNSS (RTK, INS), 3G, 4G LTE, 5G, CAT-M1, NB-IoT, Wifi Halow, WiGig, LoRa, ZigFox, ZigBee, Thread, Matter, BLE Mesh, RF-ID UHF, SDR, ASK/FSK Modem, Starlink Direct To Cell ready…
- 🏆 Best in class flexibility: Versatile and scalable through various M.2 modules, you can use a single hardware and software base for multiple projects.
- 🏆 Best in class community support: Software support, OS and frameworks compatibility…
- 🏆 Best in class AI Power Efficiency: 3 TOPS/W with the Hailo 8 TPU now and 11.4 TOPS/W expected with the Hailo-10H TPU
- Cybersecurity: Your data are protected with a TMP2.0 secure cryptoprocessor.
Austral Xplorer4 and Xplorer5 are a family of products for use when Reliability is not an option. They can be used for IIoT, AIoT & Edge Gateways or Controllers but also as Embedded Mission Computers for harsh environment.
Installed near antennas, it is intented for a Web, Remote or voice HMI and can be used with or without AI, Storage, Wireless or Positioning options. Its powerfull AI can analyze tirelessly many IP cameras at the same time. It is design to Professional, Scientific, Defence industry, Academic project, Sailor and AI Geek in :
- Industry 4.0, Energy, Oil & Gas: Machine Automation, PLC/Cloud Gateway, Overall Equipement Effectiveness, Predictive Maintenance, Smart Quality control….
- City Infrastructure Management: Parking, Lighting, Occupancy Detection, Traffic control, Pedestrian Safety, Population Flow, Air/Water/Noise quality monitoring, Waste… Building, Ship: Automation, Voice Assistant, Energy, Safety, Security, Fire Control, Climate Control, HVAC…
- Farming: Smart supervision & automation for Animals & Plants
- Store: Smart thief surveillance
- Medical, Independence Assistance: Smart supervision & automation for Hospital bed, Handicap, Physical or mental loss of autonomy…
- Sport: Electronic referee, Statistics
- Big Data Collection: Site survey, Data+Video Recorder for AI training, Ocean Science (Survey, Buoy…)
- DIY & Academic:
- Residential, Boat, RV: Smart surveillance and automation using Frigate AI/NVR + Home Assistant + Node-Red + a voice assistant
- Education: ROS2, Node-Red, TensorFlow, OpenPLC… learning
Provide intelligence and connectivity to Boats, Unmanned Vehicles (UAV, USV, AUV), Robots, Special Vehicles, Farming vehicles, Offshore Energy, Oil & Gaz systems…
- Software-Defined Vehicle: AI Detection, Obstacle Avoidance, Navigation, ML Control-Command…
- Marine Mission Computer: True Wind Calculation, Sails automation, Autopilot brain, Foiler Flight Control System, MultiFunction Display gateway…
- Consumable Military Hardware: ITAR-free Training target, Decoy, Drone…
- Voyage/Flight Data Recorder for non IMO ship or unmanned vehicle
- Tracking: Cloud gateway
- Human Attentivness Control of Driver, Pilot, Skipper…
- DIY Marine Computer: Open Plotter platform for sailors
The M.2 Hailo 8 TPU outperform onchip TPU solutions (iMX95, RK3588…) in performance and upgradability and outperform NVIDIA solutions in power efficiency :
– CM4+Hailo_8 vs Orin Nano 1/2: Inference Time 13 times lower
– CM4+Hailo 8 vs Orin Nano 2/2: Double FPS and same inference time at 2/3 of the power
– Hailo 8 vs Xavier NX : “Similar performance at a fraction of power consumption and cost”
– Hailo 8 vs AGX Xavier : “NVIDIA solution is more flexible since you can also do training on the platform, while Hailo-8 is only made for low-power inference which it does at 20 times the power efficiency”.
The new Asic based M.2 Hailo-10H TPU, expected Q4 2024, outperforms NVIDIA GPU based solution and Intel’s Core Ultra “AI Boost” NPU.
“The Hailo-10 brings generative AI to the edge, it can run Llama2-7B with up to 10 tokens per second (TPS) at under 5W of power, while it can generate one image from text in under 5 seconds using Stable Diffusion 2.1 in the same power envelope. Like other many technologies, generative AI is moving from the cloud to the device itself enabling lower latency and offline support”.
Evolution of edge AI from discriminative AI to generative AI – Source: Hailo’s blog
With its exceptional modularity (3x M.2 modules, 1x CPU module, 2x LPWAN multiprotocol modules), you can tailor the hardware to your needs with two PCB variants :
The Model E includes M.2 modules of 3 types B, E, M and a PCIE 2.0 packet switch :
- Type M 2230 is reserved for SSD.
- Type B will be used preferably for cellular communication but can be use also more SSD or advanced positionning.
- Type E includes a 2-lane PCIE to take full advantage of TPU performance but can be used also for Cellular IoT or advanced positionning.
The Model M includes M.2 modules of 2 types: 1xE and 2xM and a PCIE 3.0 packet switch to take full advantage of SSD performance and the new CM5 and Hailo10H products.
- Type M 2230 is reserved for an SSD
- Type B will be used preferably for cellular communication or advanced positionning.
- Type M 2242 for Hailo 8 or Hailo10H TPU, or more SSD
Choose or configure your Xplorer 🔎
Keep me updated, Questions, Pricing... ❓💲⏳🛒
OS and framework Compatibility 🔎
👨💼✍💻 Will be here soon 👨💼✍💻