This tool is available on the x86 Host System and NVIDIA DRIVE^™ OS Linux.

The DriveWorks Post-record Checker verifies data integrity for each recording session.

Usage

The tool is found in the tools/capture folder. Run it by executing:

./postrecord-checker --recording_session=<comma-separated list of paths>
                     --output=</path/to/output/report>
                     --sensor=<camera,radar,lidar,imu,gps,can,all>
                     --smoothing=<true/false>
                     --disable_decode=<true/false>

The postrecorder-checker does the following:

Checks recording session metadata to verify that all requried files exist and follow the naming convention.
Analyzes sensor timestamps.

The postrecorder-checker outputs a sensor timestamp delta that is calculated as timestamps difference from consecutive sensor frames. Those deltas are then used to produce a delta histogram. There are 22 bins in the histogram, each of which has a step size that is 10% of the expected timestamp delta. The first and last bin corresponds to delta values that are less than 0 or larger than 2 times of expected delta value.

By default postrecorder-checker sets --sensor=all, which means that it parses all the available sensor data. But you can also direct it to parse a subset of the sensors. For example, setting --sensor=camera,radar,lidar causes the checker to process the results for only camera, radar, and lidar sensors.

By default, postrecorder-checker sets --smoothing=true, which enables smoothing for the timestamps due to the host system time jittering.

By default, postrecorder-checker sets --disable_decode=false, which allows the tool to decode sensor frames/packets in order to extract the timestamps.

By default, postrecorder-checker will store the results within each session directory. Setting --output=</path/to/output/report> collates the results under the specified directory.

You can find python script batch_run.py in tools/capture/scripts folder to process multiple recording sessions under a parent folder

python ./batch_run.py -b <location of postrecord-checker binary> -i <input parent folder> -o <output folder to store the analysis results>

The output includes sensor timestamp files, i.e. radar.bin.ts and metadata report in report.json. To visualize timestamp delta histograms, user can use python script plot_data.py in tools folder

python ./plot_data.py -p <output folder contains the analysis results>

If user provides a single folder which contains the result produced by postrecord-checker, it plots the delta histogram inside the single folder. If user provides a parent folder which contains multiple result folders produced by batch_run script, it traverses all the folders and plots average delta histogram inside the parent folder.