Volume Estimation
One of the big opportunities in the shipping industry is remote cargo monitoring. Cargo monitoring is an important step in mitigating theft and lost shipments, but often requires sending out many people to manually open each trailer or container. Low power, native distance measurement, and grayscale imaging capabilities make flash LiDAR a fantastic technology for remote cargo monitoring.
Originally, our approach involved identifying the background and then finding the volume between background and foreground pixels. Unfortunately, a phenomenon known as multipath can cause distortion in LiDAR measurements. In the highly reflective environment inside a trailer, this is especially apparent in the distorted floor. With such a distorted background, the original approach would not work.
I was tasked with coming up with a new algorithm which could handle the distortions caused by multipath. My resulting algorithm worked by
Identifying the background image
Assigning foreground pixels to subvolumes within the trailer
Determining the occupancy status of each subvolume based on a blend of how many pixels are present and how far they are from the sensor
Any subvolume behind or below an occupied subvolume is also occupied
This approach was more robust to multipath because it didn’t depend on the accuracy of the background distance, it just used that information for filtering purposes. It also had the advantage of being able to measure free floor area, which is a metric that customers were interested in but our old algorithm couldn’t easily measure.
Here are a few videos of my algorithm in action, one from my apartment and one from the inside of a trailer.
