Timing that control engineers can trust
Effective rate, interval stability, timestamp jitter, missing periods, duplicates, non-monotonic stamps, cumulative clock drift, bursts, and sequence gaps.
Engineering diagnostic bench
Understand your motion data before trusting it.
Analyze timing, jitter, noise, drift, vibration, Allan deviation, clipping, temperature behavior, calibration readiness, and multi-IMU consistency—directly in your browser.
No account required
Professional report export
Raw data never uploaded
Start a characterizationLocal · private
Drop a recording here
Upload one sensor, several redundant IMUs, or open the included engineering example.
CSV / TXT native
MCAP native browser reader
ROS 1 BAG native browser reader
ROS 2 FOLDER DB3 + MCAP splits
High-rate awareStreaming parsing, bounded analysis buffers, decimated plots.
Technically honestQuality and stability are separated from ground-truth accuracy.
A virtual sensor laboratory, in one page.
From first-pass integrity checks to timing diagnostics, vibration analysis, stochastic noise characterization, temperature dependence, calibration output, and multi-sensor comparison.
No dataset loadedUpload or open example
—Duration
—Samples
—Effective rate
—Data integrity
Compute engineBrowser CPU · cooperative chunks
ParserCSV native · ROS loaders on demand
NumericsFloat64 statistics · bounded plots
Privacy boundaryRaw recording remains local
Ready for a recording
Use your own CSV or open the example dataset to explore the complete analysis workspace.
Acquisition provenance & reproducibility
Inspect exactly what was parsed: recording segments, storage backend, selected ROS topic, serialization, timestamp source, units, local compute path, and analysis decimation. This information is carried into the report and machine-readable manifest.
Local file accessSchema verified
Incremental statisticsFloat64 timingBounded min/max plotsWelch PSDOverlapping Allan deviationTransparent confidence checks
Source & storage
Files, backend, size, and split-bag structure
Topic & message contract
ROS type, serialization, frame, and timestamps
Analysis engine
Local resources and numerical method
Six-axis time trace
Common-time preview in SI units; offset removed for visualization
Visualization is decimated; statistics use every parsed sample.
SIIntegrity event timeline
Missing periods, timestamp anomalies, clipping, and discontinuity candidates
Recording segments
ROS 2 split files are ordered from metadata.yaml when available, then merged into one logical timeline
CSV / delimited textStreaming parser, automatic mapping, SI normalization, large-file bounded memory.
ROS 2 SQLite3 foldermetadata.yaml + one or many .db3 segments; native sensor_msgs/msg/Imu CDR; each segment opens locally in WASM memory.
MCAP / ROS 2 MCAP folderIndexed or sequential MCAP, compressed chunks, embedded ROS 1/ROS 2 schemas, JSON channels.
ROS 1 .bagTopic discovery and embedded message definitions; uncompressed, LZ4, and BZ2 handlers loaded on demand.
Motion overview
Acceleration and angular-rate magnitude over time
Acceleration magnitudeAngular-rate magnitude
Main findings
Prioritized engineering observations
Recording summary
Observed properties and acquisition health
Automatic motion segmentation
Stationary and dynamic regions detected from the data
Mean sample rate—From complete recording duration
RMS jitter—Relative to nominal period
Dropped samples—Timestamp + sequence estimate
Clock drift—End-to-end cumulative drift
Sample interval over time
Observed Δt versus nominal period
Jitter distribution
Histogram of interval error
Timing diagnostics
Percentiles, anomalies, and timestamp quality
Interpretation
What the timing trace means for a control system
Instantaneous sample-rate stability
Robust rolling median rate; scheduler and batching patterns become visible
Rate is derived from positive source intervals after compensating plot stride.
fₛ(t)=1/ΔtCumulative timing phase error
Measured acquisition timeline versus nominal clock
Separates random jitter from systematic sample-clock mismatch.
eₜ(k)=tₖ−t₀−kT₀Measurement-to-record timestamp delta
Median-centered transport / buffering trace when both ROS header and storage times exist
Absolute latency is meaningful only when both clocks share a synchronized domain.
δLᵢ=(tᵣ−tₘ)ᵢ−median(tᵣ−tₘ)Transport timing diagnostics
Raw timestamp delta, jitter percentiles, span, and interpretation boundary
No paired timestamps were found.
ROS header and record times are both required.
Detected timing events
Largest gaps, duplicate/backward stamps, and high-jitter intervals
| Time | Event | Observed interval | Nominal periods | Severity |
|---|
Stationary noise, bias, and distribution statistics
Metrics are flagged when motion contaminates a stationary estimate
| Channel | Mean / bias | Std. deviation | RMS | Peak-to-peak | Skewness | Kurtosis |
|---|
Channel distribution
Robust histogram around the median
Gravity sanity check
Useful for stationary scale and calibration checks
Mean measured gravity—
Deviation from standard gravity—
Gravity magnitude noise—
Rolling bias & local noise
Windowed mean and standard deviation reveal warm-up, drift, and nonstationarity
Adaptive window
μᵥ(t), σᵥ(t)Autocorrelation
Identifies colored noise, periodicity, and effective correlation time
Computed on the selected stationary channel after mean removal.
ρ(k)=R(k)/R(0)Cross-axis correlation matrix
Common-mode vibration, electrical coupling, and axis dependence
Distribution diagnostics
Robust normality indicators and tail behavior
Select a channel to calculate distribution diagnostics.
Frequency & vibration lab
Welch power spectral density with dominant peak and alias-risk detection
Power spectral density
One-sided PSD, Hann window, overlapping Welch averages
Dominant components
Ranked non-DC spectral peaks
| Frequency | Relative level | Assessment |
|---|
Time–frequency spectrogram
Short-time Fourier transform for transient vibration and changing resonances
Hann windows, 50% overlap, log-power display.
S(t,f)=|STFT{x}|²Cumulative integrated RMS
How total noise or vibration accumulates with bandwidth
Integrated from DC to Nyquist.
RMS(B)=√∫₀ᴮ PSD(f)dfMagnitude-squared coherence
Frequency-dependent coupling between two axes
Spectral metrics
Bandwidth-aware quantitative summary
Broadband noise density—
Integrated RMS to Nyquist—
Equivalent noise bandwidth—
Allan deviation & stochastic noise
Overlapping Allan deviation with dataset-sufficiency checks
Allan deviation
Logarithmic averaging time and deviation
Random walk estimate
—
—
Bias instability estimate
—
—
All-axis Allan overlay
Compare stochastic behavior and common slope regions across the sensor triad
Slope-region diagnostics
Local log–log slopes and coefficient confidence
| Channel | Short-term slope | Minimum σ | τ at minimum | Confidence |
|---|
Bias versus temperature
Linear dependency preview; polynomial models belong in a dedicated temperature sweep
Temperature model
Observed dependency in this recording
Temperature span—
Linear slope—
Coefficient of determination—
Temperature-model residual
Remaining bias after removing the fitted linear dependency
Warm-up and temporal stability
Temperature and selected channel versus elapsed time
Warm-up estimate requires a meaningful temperature span.
b(t), T(t)Zero-rate gyroscope estimate
Mean vector during detected stationary samples
Accelerometer sanity estimate
Gravity magnitude and apparent scale residual
Guided six-position calibration
Upload labeled +X, −X, +Y, −Y, +Z, −Z stationary captures for full bias, scale, misalignment, and non-orthogonality estimation
+X upWaiting for labeled capture
−X upWaiting for labeled capture
+Y upWaiting for labeled capture
−Y upWaiting for labeled capture
+Z upWaiting for labeled capture
−Z upWaiting for labeled capture
Multi-IMU comparison
Compare two or three recordings after unit normalization and common-frame preparation
| Recording | Rate | RMS jitter | Gyro RMS | Accel RMS | Dropped | Assessment |
|---|
Comparison interpretation
Potential outliers and shared failure modes
Fusion simulation
Mean, variance weighting, median, and residual rejection
Numerical fusion comparison is shown only as an engineering preview. Formal conclusions require synchronized recordings, known mounting rotations, equal bandwidth, and a shared body frame.
Pairwise residual & fusion study
Signals are interpolated to a common elapsed-time grid
Pairwise residuals
A−B, A−C, and B−C after common-grid interpolation
Fusion noise comparison
Observed standard deviation under explicit assumptions
Olive Robotics · Engineering characterization
IMU Data Quality Report
A technically defensible summary of observed timing, integrity, noise, vibration, stability, and calibration-readiness properties.
Measurement boundary: this report characterizes the uploaded data’s quality, stability, timing, repeatability, and consistency. Absolute motion or orientation accuracy requires synchronized ground-truth data.
Noise and long-term stability
Bias, RMS noise, robust distributions, gravity checks, Allan deviation, random walk, bias-instability estimates, and reliability warnings when data is insufficient.
Frequency and vibration insight
FFT/Welch PSD, dominant peaks, harmonics, motor-RPM correlation, resonance clues, aliasing risk, and axis-dependent vibration behavior.
Temperature behavior
Bias versus temperature, warm-up behavior, regression quality, compensation-model preparation, and explicit warnings when normal motion contaminates thermal estimates.
Calibration-ready output
Zero-rate bias estimates, gravity scaling checks, six-position workflow scaffolding, and exports designed for ROS 2 YAML or embedded firmware integration.
Multi-redundant IMU comparison
Normalize recordings, compare bias/noise/timing, expose outliers, and prepare for timestamp alignment, mounting transforms, residuals, fault isolation, and robust fusion studies.
Your sensor data stays on your machine.
File reading, message decoding, statistics, spectral calculations, visualizations, and report generation run inside the browser. Raw recordings are never uploaded. On first use of MCAP or ROS bags, pinned open-source parser modules may be fetched and cached by the browser; they execute locally against the selected files.
Confirm data mapping
Review detected fields, units, sensor range, and test context.
Timestamp and optional sequence
Accelerometer channels
Gyroscope channels
Metadata and analysis context
Range and custom scaling
Why this matters: data filtered to 20 Hz should not be compared directly with data at 400 Hz bandwidth, and dynamic robot motion should not be reported as stationary sensor noise. The report records this context.
Analysis begins locally after you confirm this mapping.
Configure ROS / MCAP recording
Select the message topic, timestamp source, acquisition context, and sensor limits.
Recording—
Storage—
Segments—
Total size—
ROS topics and time base
Standard sensor_msgs/Imu is decoded natively. MCAP embedded schemas also support custom ROS messages.
Nearest/interpolated sensor_msgs/Temperature samples are aligned locally.
Characterization context
Custom-object field paths
Example: imu.accel or linearAcceleration
Example: imu.gyro or angularVelocity
Used when temperature is in the same message.
Local parser: inspecting the recording and embedded schema.
Only parser code may be fetched; the selected recording stays inside this browser tab.
Ready.
The information, specifications, and materials provided on this website are published by Olive Robotics GmbH for general informational and commercial purposes only. While every effort is made to ensure accuracy and timeliness, all specifications, performance figures, availability details, and product roadmaps remain non-binding and subject to modification, withdrawal, or correction without prior notice. Nothing on this website shall be construed as constituting a warranted characteristic within the meaning of §§ 434 et seq. BGB, nor as a binding commercial offer within the meaning of § 145 BGB, unless expressly confirmed in writing by Olive Robotics GmbH.
Access to and use of this website is at the user’s sole risk. Olive Robotics GmbH disclaims, to the maximum extent permitted by applicable law, all liability for direct, indirect, incidental, or consequential damages arising from the use or misuse of this website, its content, or any linked external resources. Pursuant to §§ 7–10 TMG, Olive Robotics GmbH is not responsible for third-party content accessible via external hyperlinks and assumes no responsibility for the availability, accuracy, or legality of such content.
All trademarks, logos, product identifiers, and intellectual property referenced on this site, including but not limited to olive®, olixSense™, olixAI™, olixOS™, and related marks, are the exclusive property of Olive Robotics GmbH. Unauthorized reproduction, modification, or commercial use is strictly prohibited. References to third-party technologies or marks are for descriptive purposes only and imply no endorsement or affiliation.
All quotations, orders, and deliveries are governed exclusively by Olive Robotics GmbH’s General Terms and Conditions (AGB) as published on this website. For consumers, statutory rights under the German Civil Code (BGB) remain unaffected.
Olive Robotics GmbH
Embedded Physical AI for Robots.