<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>HOLON Research</title><description>Reports, benchmarks and buyer guides on capture hardware for robot learning, wearables and edge vision.</description><link>https://www.holonai.ai/</link><item><title>Compute for Capture Devices and Edge Boxes: RK3576, RK3588, Allwinner A733, Raspberry Pi 5 and Jetson Orin</title><link>https://www.holonai.ai/research/capture-head-compute-rk3576-rk3588-jetson/</link><guid isPermaLink="true">https://www.holonai.ai/research/capture-head-compute-rk3576-rk3588-jetson/</guid><description>For a recorder that only acquires, timestamps and stores, a low-cost Rockchip or Allwinner board is enough. For six raw cameras or on-device models, RK3588 or RK3576 (6 TOPS NPU) covers mainstream detection at low power; Jetson Orin is the choice when you need CUDA, larger models or the GMSL camera ecosystem. Benchmark your own pipeline, not TOPS.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>RK3576 vs RK3588</category><category>RK3588 vs Jetson Orin</category><category>edge AI box Rockchip</category><category>SBC for stereo camera recording</category><category>Jetson Orin GMSL camera</category><category>compute for egocentric headset</category></item><item><title>Edge AI Boxes for Retail and Inspection: A Hardware Checklist That Survives the Pilot</title><link>https://www.holonai.ai/research/edge-ai-box-retail-analytics/</link><guid isPermaLink="true">https://www.holonai.ai/research/edge-ai-box-retail-analytics/</guid><description>Edge-vision pilots usually fail on hardware operations, not models: throttling in sealed enclosures, camera streams that drop frames, and no update or recovery path. Choose the box by measured FPS and latency for your model at maximum ambient temperature, keep identifying data on the device, and design remote updates and health telemetry before the first store.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>edge AI box retail analytics</category><category>RK3588 edge AI box</category><category>Jetson retail analytics hardware</category><category>privacy preserving retail camera</category><category>edge AI deployment checklist</category><category>Hailo vs Jetson vs Rockchip</category></item><item><title>Egocentric Data-Capture Hardware: 2026 Buyer&apos;s Benchmark</title><link>https://www.holonai.ai/research/egocentric-capture-hardware-2026/</link><guid isPermaLink="true">https://www.holonai.ai/research/egocentric-capture-hardware-2026/</guid><description>Egocentric capture hardware for robot learning falls into five classes: phone rigs, stereo RGB-IMU headsets, four-to-six-camera SLAM headsets, modular ego-plus-wrist-plus-gripper kits, and custom builds. Most teams should start from an existing SLAM-class headset and pay for targeted changes — frame rate, IMU rate, exposure timestamps, raw output — rather than a ground-up build.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>egocentric data collection hardware</category><category>egocentric camera headset robotics</category><category>robot data collection headset supplier</category><category>head mounted camera for robot learning</category><category>VLA data collection device</category><category>egocentric video dataset hardware China</category></item><item><title>Global-Shutter Sensors for Robot Tracking and Egocentric Capture: OV9281, AR0234, IMX296 and Alternatives</title><link>https://www.holonai.ai/research/global-shutter-sensors-robotics/</link><guid isPermaLink="true">https://www.holonai.ai/research/global-shutter-sensors-robotics/</guid><description>For monochrome tracking cameras, the OV9281/OV9282 (1280×800, 3 µm, up to 120 fps) remain the default; OV2311 adds 2 MP with enhanced near-infrared. For egocentric RGB, AR0234 (1920×1200, 2.3 MP, global shutter) is the common choice, with IMX296 and OG02B10 as alternatives. The module, lens and trigger path matter more than the sensor.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>OV9281 vs AR0234</category><category>global shutter camera for robotics</category><category>global shutter vs rolling shutter robot</category><category>OV9281 camera module trigger</category><category>AR0234 camera module 60fps</category><category>850nm fisheye lens tracking camera</category></item><item><title>The Hardware RFP Clauses That Decide Who Owns Your Product: Firmware, Data, Calibration and Design Files</title><link>https://www.holonai.ai/research/hardware-rfp-clauses-firmware-data-ownership/</link><guid isPermaLink="true">https://www.holonai.ai/research/hardware-rfp-clauses-firmware-data-ownership/</guid><description>Before the first sample, a hardware RFP should require: all device data routable to a server you control; complete firmware source at delivery or in escrow; per-unit calibration files in an open format; a full BOM with manufacturer part numbers; design-file transfer terms; NRE priced separately from unit cost; and a measured acceptance test that gates payment.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>hardware RFP template</category><category>firmware source code escrow manufacturer</category><category>OEM contract data ownership</category><category>Shenzhen manufacturer contract clauses</category><category>white label device data ownership</category><category>hardware startup vendor agreement</category></item><item><title>Holon Labs: Event Cameras for Ultrafast Defect Inspection</title><link>https://www.holonai.ai/research/lab-event-camera-defect-inspection/</link><guid isPermaLink="true">https://www.holonai.ai/research/lab-event-camera-defect-inspection/</guid><description>Event cameras report per-pixel brightness changes asynchronously with microsecond-scale latency instead of full frames, so fast-moving or vibrating defects that blur in a frame camera stay sharp in event data. Combined with computational reconstruction, they make it practical to inspect dynamic processes at speeds and data rates frame cameras cannot match.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>event camera defect inspection</category><category>neuromorphic imaging industrial inspection</category><category>event-based vision robotics</category><category>dynamic vision sensor inspection</category><category>high speed defect detection camera</category></item><item><title>Holon Labs: Neuromorphic Imaging for Tracking Around Corners</title><link>https://www.holonai.ai/research/lab-neuromorphic-tracking-nlos/</link><guid isPermaLink="true">https://www.holonai.ai/research/lab-neuromorphic-tracking-nlos/</guid><description>Non-line-of-sight tracking infers where a hidden object is by analysing light it scatters onto visible surfaces. Event cameras make this efficient because they record only the small changes in that scattered light, cutting data and latency compared with frame-based approaches.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>non line of sight tracking event camera</category><category>neuromorphic imaging tracking</category><category>NLOS imaging robotics</category><category>seeing around corners camera</category></item><item><title>MIPI, GMSL2, USB3 or 10GbE? Choosing the Camera Link for Multi-Camera Robots</title><link>https://www.holonai.ai/research/mipi-gmsl-usb3-10gbe-multicamera/</link><guid isPermaLink="true">https://www.holonai.ai/research/mipi-gmsl-usb3-10gbe-multicamera/</guid><description>Use MIPI CSI-2 when cameras sit within about 30 cm of the processor; GMSL2 or GMSL3 when they are up to 15 m away on a moving robot and you need power, control and trigger on one coax; USB3 for quick prototypes under about 3 Gbps total; and 10GbE when a wearable must send raw multi-camera streams to a backpack or base computer.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>GMSL2 camera Jetson Orin</category><category>MIPI to GMSL converter</category><category>GMSL vs USB3 camera</category><category>multi camera bandwidth calculation</category><category>FPD-Link vs GMSL</category><category>camera link for robots</category><category>MAX96717 MAX96712</category></item><item><title>Can a Pet Collar Measure Heart and Breathing Rate? An Engineering Evidence Plan</title><link>https://www.holonai.ai/research/pet-collar-heart-respiration-sensing/</link><guid isPermaLink="true">https://www.holonai.ai/research/pet-collar-heart-respiration-sensing/</guid><description>A collar can plausibly estimate resting heart rate and respiratory rate from mechanical signals when the animal is still, with signal-quality gating and animal-specific validation. It cannot measure core temperature, blood oxygen or blood pressure on its own. Claims should be limited to what a synchronised reference has verified, and the device must report &apos;insufficient signal&apos; instead of guessing.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>pet collar heart rate accuracy</category><category>dog collar respiratory rate sensor</category><category>ballistocardiography pet</category><category>smart collar vital signs</category><category>pet wearable validation</category><category>BCG sensor collar</category></item><item><title>How to Read a Shenzhen Camera-Module Quote (and the Eight Questions That Expose a Bad One)</title><link>https://www.holonai.ai/research/reading-shenzhen-camera-module-quotes/</link><guid isPermaLink="true">https://www.holonai.ai/research/reading-shenzhen-camera-module-quotes/</guid><description>Treat every camera-module quote as a hypothesis. Confirm the exact sensor, the delivered resolution, frame rate and format per interface, whether a trigger pin is exposed, IMU part and rate, measured distortion for the actual lens, per-unit calibration, SDK status and what the price includes. In our 2026 reviews, most datasheets contradicted themselves on at least one of these.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>camera module quote China</category><category>Shenzhen camera module supplier</category><category>stereo camera module datasheet</category><category>AR0234 stereo USB camera module</category><category>camera module MOQ price</category><category>how to source camera modules from China</category></item><item><title>Field Notes: Running Egocentric Data Collection Across Southeast Asia</title><link>https://www.holonai.ai/research/sea-egocentric-data-operations/</link><guid isPermaLink="true">https://www.holonai.ai/research/sea-egocentric-data-operations/</guid><description>Workplace egocentric collection succeeds or fails on operations: employer consent before the first session, trained participants, devices that are charged, clean and correctly mounted, protocol-specific QA that catches duplicates and filler footage, and reliable upload. Hardware choice matters, but process discipline decides yield.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>egocentric data collection Southeast Asia</category><category>data collection Malaysia robotics</category><category>data collection Indonesia AI</category><category>data collection Vietnam Thailand Philippines</category><category>workplace egocentric video collection</category><category>robot learning data operations</category></item><item><title>The Shenzhen Factory-Visit Playbook for Hardware Founders</title><link>https://www.holonai.ai/research/shenzhen-factory-visit-playbook/</link><guid isPermaLink="true">https://www.holonai.ai/research/shenzhen-factory-visit-playbook/</guid><description>A productive factory trip is decided before you land: pre-qualify suppliers against a written spec, send the agenda and questions in Chinese a week ahead, see samples and test equipment rather than showrooms, debrief each evening, and leave with a written comparison and next steps per supplier.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>Shenzhen factory visit</category><category>visiting suppliers in Shenzhen</category><category>Dongguan factory audit hardware startup</category><category>how to visit Chinese factories</category><category>hardware sourcing trip China</category></item><item><title>Timing Is the Spec: Synchronisation Requirements for Robot-Learning Capture Hardware</title><link>https://www.holonai.ai/research/timing-is-the-spec/</link><guid isPermaLink="true">https://www.holonai.ai/research/timing-is-the-spec/</guid><description>For robot-learning capture, &apos;synchronised&apos; must mean that every sample carries the time its light or motion was actually acquired, on one shared clock. Practical targets used by frontier teams are under 100 µs between tracking cameras, under 500 µs camera-to-IMU, and under 1 ms for RGB and radio events — verified by measurement, not vendor statements.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>multi-camera synchronization</category><category>camera IMU synchronization</category><category>exposure timestamp</category><category>hardware trigger</category><category>PTP camera sync</category><category>egocentric data collection hardware</category><category>robot learning data quality</category></item><item><title>Body and Object Trackers for Robot-Learning Data: LED Pucks, Self-Tracking Pucks, IMU Suits and Custom Designs</title><link>https://www.holonai.ai/research/trackers-for-robot-learning/</link><guid isPermaLink="true">https://www.holonai.ai/research/trackers-for-robot-learning/</guid><description>Trackers for robot-learning data use one of four architectures: IR-LED constellations seen by headset cameras, self-tracking camera pucks, IMU-only suits, or hybrids with UWB ranging. Off-the-shelf options are fine for testing, but most cap tracker count, compute pose on-device and lack raw, synchronised data — the requirements a training pipeline needs.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>IR LED body tracker alternative</category><category>body tracker robot learning</category><category>self-tracking puck robotics</category><category>full body tracking data collection</category><category>IR LED tracker puck design</category><category>UWB tracker headset synchronization</category></item><item><title>White-Label or Custom? A Dual-Track Strategy for Funded Wearable Startups</title><link>https://www.holonai.ai/research/white-label-vs-custom-wearables/</link><guid isPermaLink="true">https://www.holonai.ai/research/white-label-vs-custom-wearables/</guid><description>Run both. A white-label device validates demand and the app in roughly 4–8 weeks; a custom design, typically 3–4 months to prototype, builds ownership and margin in parallel. Switch volume to the custom track once it passes validation. In both tracks, insist that device data can flow to your own cloud.</description><pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate><category>white label wearable manufacturer</category><category>custom wearable development Shenzhen</category><category>pet collar manufacturer</category><category>smart ring OEM</category><category>wearable startup hardware strategy</category><category>ODM vs custom hardware</category></item></channel></rss>