Who should use this
Robot learning, VLA, imitation learning, teleoperation and embodied AI laboratories building multimodal data capture rigs.
Sensing, Vision & Data Capture
Prepare camera, LiDAR, joint-state, telemetry, trigger, timestamping, host bandwidth and storage inputs for multimodal robot data capture reviews.
Robot learning, VLA, imitation learning, teleoperation and embodied AI laboratories building multimodal data capture rigs.
The guide defines capture interfaces and hardware timing requirements while keeping dataset, model and latency validation outside the sourcing scope.
Bring GMSL2 or USB3 camera channels, LiDAR or depth streams, joint-state and telemetry streams, frame rates, point-cloud rates, trigger requirements, host input bandwidth, NVMe logging and export format.
When to use this guide
Key engineering decisions
List GMSL2 or USB3 camera channels, LiDAR, depth, joint-state and master-slave telemetry streams with source rates.
PTP / IEEE 1588, PPS when supplied, hardware trigger and timestamping requirements can be captured as customer-supplied review inputs when multimodal data streams must be aligned.
Estimate host input bandwidth, bus allocation, NVMe throughput, recording duration and available storage margin.
Identify requested data export format and capture ownership without implying dataset suitability or model-training results.
RFQ preparation checklist
Engineering file considerations
Not covered by this guide
Use the guide checklist to submit one RFQ with product, file and subsystem context. Controlled access may apply to STEP/STP, native CAD, full drawings and supplier-sensitive records.