Python device driver and point cloud processing software for parallel data collection and processing with Livox Mid-70 LiDAR sensor. Intended for use in remote snow measurement using a Raspberry Pi. Developed for NCAR|UCAR.
- Tested on Raspberry Pi 3,4 with standard Raspian (Raspberry Pi OS). Python 3.8+ required.
- For GPS timestamp data labelling, Adafruit Ultimate GPS v3 module
- Developed using the Livox Mid-70. See https://www.livoxtech.com/mid-70/downloads for LiDAR User Manual
A corresponding Raspberry Pi OS has been made to make initial steps simple. If this image is not used directly, a few steps must be taken to configure the Raspberry Pi:
- Set a static IP address of 192.168.1.2 for the Ethernet port (symlink eth0).
- Allow access to the hardware UART pins by disabling the serial console in the raspi-config menu.
- Upgrade Python to version 3.8 or higher and download all necessary libraries (check module imports for non Standard Library modules).
Due to the large memory requirements of point cloud data, a high storage USB flash drive or external hard drive is likely necessary if data is not significantly downsized. In openpylivox.py, pointcloudprocessor.py, and SnowMeasureLivox.py, the path string for saving data must be changed to reflect the external storage.
All parameters related to scheduling and data processing are found in main_config.ini and processing_config.ini respectively. Due to the nature of the Ground Elevation routine, arrays must be limited to identical bin size and number of bins for ground level subtraction and direct comparison. Read the comments in those .ini files and their respective use in SnowMeasureLivox.py and pointcloudprocessor.py.
Collection sessions are determined by the number of recordings and the duration of recordings. An entire session is performed when SnowMeasureLivox.py is executed; to perform another collection session, an OS level scheduler (such as a cron job) or altering SnowMeasureLivox.py to collect again is required.
Notes:
- A ground height measurement routine has not been fully developed. It should be as simple as altering a copy of SnowMeasureLivox.py to perform a collection session, and passing 0 as a placeholder for ground_truth_elevations (~line 60-70) in a copy of pointcloudprocessor.py. For a remote deployment, periodic measurements of ground elevation can be scheduled into the recording schedule, or can be set to run on boot.
- There are print statements scattered throughout all modules used to verify the multiprocessing functionality during development. After testing and verification, these can be deleted to reduce overhead, or changed to logging statements and recorded to a .log file if post-hoc debugging is desired.
- If the raw point cloud data is not desired, they should be deleted using os/shutil or something similar after all collections are complete in SnowMeasureLivox.py
This project is still in development, please be patient!