Over the last few days, we have been making good progress with our LiDAR endeavors. Neil and Charlie have been working on weight-efficient methods to attach the LiDAR to our drone, Kia. The initial prototype was too heavy, and a test run over-heated Kia’s battery. It turns out that we can cover one of Kia’s sensors without catastrophe, which will allow us to attach the LiDAR to Kia underneath said sensor without needing extra material like a platform for the LiDAR to sit on.
It will be useful for us to sync up the LiDAR data with GPS data from Kia. Fortunately, both the LiDAR and Kia transmit time-data. Unfortunately, that data is not easily accessible, much less synchronizable. Kellan and Nic have been working on methods to extract this data so we can sync up and get going. They have also been working on installing software and writing code for analyzing point clouds (3-D depictions of LiDAR readings).
I have been familiarizing myself with the tools and methods used for analyzing point clouds. Once we have gathered our LiDAR data, how do we extract meaning from any of it? First we will need to reduce noise, which can be done by using statistical analysis to detect outlier points (laser reflects off of a piece of airborne dust, for instance). Since we will be looking for particular kinds of things (archaeological ruins, bird nests, etc.), we will then need to have methods for detecting particular features. For example, the foundation of a structure will likely have near-straight lines. If we have a tool that will highlight all near-straight lines, we can search for archaeological ruins more efficiently. There are many object-detection techniques, all of which involve very clever math. If we’re lucky, any noise-filtering and object-detection algorithms that we may need are already out there. If not, enough work has been done in this field that synthesizing existing methods will (hopefully) be doable.
We are currently searching for a renegade tablet that seems to have wandered off. I can’t imagine it can run very fast or far, so we should come across it soon if it doesn’t get tired and come home by itself.