LiDAR technology plays a vital role in capturing the economic importance of forest and in policy making through monitoring forests and agricultural resources remotely. It is a prominent remote sensing technology to capture 3-D forest structure with spatial and temporal meta-data such as coordinates, colors, GPS locations, and timestamps. This project aims to use LiDAR data to recreate forest structure in 3-Dimensions to perform forest inventory analysis and object characterizations. This study also presents an automated pipeline of methodologies to identify individual trees from forest vegetation and estimate tree characteristics such as canopy heights and trunk diameter. The derived forest structure is used as an arena to simulate animal movements using Deep Q Reinforcement Learning in 2-Dimensions. Bats are chosen as the model group as they are sensitive to the density of vegetation. This paper proposes a work ow for any species of bat to take appropriate action depending on the environment by evaluating the rewards for each available action. This paper also describes a simulation-based investigation of the behavior of the movements in a controlled environment to evaluate decision-making strategies, the performance of collision avoidance, and the success rate of traveling across the forest.
pip install laspy
pip install pptk
pip install pye57 conda install xerces-c
Python Version: > 3.0