One continuous real-world task recording formed the source capture evaluated in this field note.
Technical Field Note #01
Measured Video + IMU Capture Characteristics
12-Segment Evaluation from a Continuous Real-World Field Recording
These measurements represent observed capture-pipeline performance across 12 consecutive segments generated from a single continuous recording under the tested configuration. They demonstrate sustained measurement consistency within this recording, not 12 independent experimental replications and not a guaranteed hardware specification. Hardware frame-level video–IMU synchronization is not claimed.
12
Consecutive measured capture segments
409.65–410.14 Hz
Observed native accelerometer effective rate
409.94–410.27 Hz
Observed native gyroscope effective rate
0 / 0
Timestamp monotonicity violations / duplicate timestamps observed
Measurement Scope
What Was Evaluated
A continuous real-world agricultural produce sorting and portioning task was recorded in South Korea. During server-side handling, the continuous recording was represented as 12 consecutive capture segments. Each segment was evaluated independently from its recorded timestamps.
The continuous recording was represented as consecutive capture units during server-side handling.
Twelve consecutive segments were measured independently from their recorded timestamps.
Video timing, native and application-level IMU, timestamp integrity, and temporal coverage were evaluated.
Public Release Context
Capture-context wording on this page has been prepared specifically for public release. Internal operational metadata is retained separately and is not reproduced here.
Methodology
Measurements Recomputed from Recorded Timestamps
Effective sensor rate was calculated from sample count and the timestamp span of each measured segment rather than from configured or requested sampling-rate values.
Effective Rate Calculation
effective_hz = (sample_count - 1) /
((last_timestamp_ns - first_timestamp_ns) / 1e9)
Timestamped accelerometer and gyroscope records parsed from
imu_native_v1.bin.
Timestamped accelerometer and gyroscope records parsed from
imu_metadata_v1.json.
Effective video FPS was derived from decoded frame count and container duration. Timing gaps were evaluated separately.
Native IMU timestamp ranges were compared with corresponding video start and end monotonic anchors in the capture-session clock domain.
Synchronization Boundary
Temporal coverage evaluates whether the IMU timestamp range brackets the corresponding video interval. It does not establish frame-level, sub-frame, exposure-level, or hardware-triggered synchronization.
Measured Results
Observed Characteristics Across the 12 Segments
29.83–30.00
Effective Video FPS
409.65–410.14 Hz
Native Accelerometer
409.94–410.27 Hz
Native Gyroscope
12 / 12
Video–IMU Temporal Bracket Coverage
Detailed Measurement Table
Minimum, median, mean, and maximum values below were computed across the 12 consecutive measured segments.
| Measurement | Minimum | Median | Mean | Maximum |
|---|---|---|---|---|
| Effective video FPS | 29.8334 | 29.9837 | 29.9646 | 30.0004 |
| Video frame max gap | 66.63 ms | 66.65 ms | 102.74 ms | 499.80 ms |
| Native accelerometer rate | 409.6461 Hz | 409.8665 Hz | 409.8693 Hz | 410.1447 Hz |
| Native gyroscope rate | 409.9429 Hz | 410.1112 Hz | 410.1097 Hz | 410.2659 Hz |
| Native accelerometer max gap | 25.98 ms | 28.02 ms | 27.50 ms | 30.46 ms |
| Native gyroscope max gap | 4.93 ms | 7.34 ms | 10.94 ms | 24.68 ms |
| Application-level accelerometer rate | 97.7867 Hz | 99.0855 Hz | 99.1258 Hz | 100.7711 Hz |
| Application-level gyroscope rate | 97.7937 Hz | 99.2673 Hz | 99.1824 Hz | 100.6345 Hz |
| Application-level accelerometer max gap | 163.80 ms | 205.78 ms | 280.85 ms | 589.34 ms |
| Application-level gyroscope max gap | 103.65 ms | 207.00 ms | 276.75 ms | 599.08 ms |
| Native IMU pre-roll | 303.17 ms | 422.30 ms | 429.58 ms | 562.22 ms |
| Native IMU post-roll | 87.40 ms | 114.37 ms | 111.53 ms | 135.30 ms |
Timestamp Integrity & Coverage
Observed Timestamp and Temporal-Coverage Results
0
Timestamp Monotonicity Violations
0
Duplicate Timestamps Observed
12 / 12
Video–IMU Bracket Coverage
Not Claimed
Hardware Frame-Level Synchronization
Detailed Evidence
- Segments evaluated
- 12 consecutive segments from one continuous field recording
- Timestamp monotonicity
- 0 violations observed across the evaluated timestamp streams
- Duplicate timestamps
- 0 duplicates observed
- Video–IMU bracket coverage
- 12 / 12 segments
- Native accelerometer
- 409.6461–410.1447 Hz observed effective rate
- Native gyroscope
- 409.9429–410.2659 Hz observed effective rate
- Effective video FPS
- 29.8334–30.0004 FPS
- Hardware frame-level sync
- Not claimed
Across the evaluated capture, the native IMU timestamp range fully bracketed the corresponding video interval in all 12 segments. This supports capture-session temporal coverage using the evaluated monotonic timestamp lineage. It does not demonstrate exposure-level, frame-accurate, sub-frame, or hardware-triggered synchronization.
Observed Limitations
Timing Gaps Are Reported, Not Hidden
499.80 ms Maximum Gap
One of the 12 evaluated segments contained a video frame interval exceeding 100 ms. The observed maximum gap was 499.80 ms. The cause was not established by this measurement.
Intervals Above 100 ms
Application-level accelerometer and gyroscope streams showed intervals exceeding 100 ms in all 12 evaluated segments. Maximum observed gaps reached 589.34 ms and 599.08 ms respectively.
Cause Not Inferred
Native timestamp streams did not show gaps of comparable magnitude. No cause is assigned to the difference between native and application-level representations.
Measurement Boundary
These results characterize one tested capture configuration and one continuous field recording. They are not a universal device guarantee, a production SLA, or evidence of independent experimental replication.
Technical Boundaries
What This Evidence Does Not Claim
NO CLAIM
Hardware frame-level video–IMU synchronization.
NO CLAIM
Sub-frame or camera-exposure timestamp synchronization.
NO CLAIM
A guaranteed 410 Hz hardware specification.
NO CLAIM
Guaranteed continuous 100 Hz application-level delivery.
NO CLAIM
The 12 segments are independent experimental sessions.
NO CLAIM
An inferred cause for observed video or application-level timing gaps.
Machine-Readable Summary
Key Technical Facts
12
Consecutive Capture Segments Evaluated
~410 Hz
Observed Native 6-DoF IMU Effective Rate
0 / 0
Monotonicity Violations / Duplicate Timestamps
12 / 12
Video–IMU Temporal Bracket Coverage
Evidence Summary
Origin Data Lab measured native and application-level 6-DoF IMU sensor streams across 12 consecutive capture segments generated from a single continuous real-world field recording. Native sensor effective rates were observed at 409.65–410.14 Hz for the accelerometer and 409.94–410.27 Hz for the gyroscope.
Across the evaluated timestamp streams, zero timestamp monotonicity violations and zero duplicate timestamps were observed, and the native IMU timestamp range fully bracketed the corresponding video recording interval in all 12 segments.
These results describe sustained measurement consistency across one continuous recording, not independent experimental replication, and do not constitute a guaranteed hardware specification or hardware frame-level synchronization claim.
Technical Validation
Need Technical Evidence for a Real-World Capture Program?
Origin Data Lab can define capture architecture, field operations, sensor requirements, metadata, validation criteria, QC, and structured delivery around project-specific Physical AI and robotics requirements.