Design and Integration of Semantic Mapping System for Forest Fire Mitigation

Abstract

Remote sensing technologies can provide an automated approach to monitor and analyze conditions in the forest environment over a period of time for forest maintenance and wildfire mitigation efforts. In particular, unmanned aerial vehicles (UAVs) are a promising remote sensing modality since they can traverse uneven terrain and provide on-demand high-resolution surveys of the environment with various sensors.

In this work, we present the mechanical design and system integration of a multi-sensing payload to harness such capabilities of UAVs and collect meaningful data in forest environments with the ultimate aim of localizing clusters of flammable vegetation. Our payload primarily consists of LiDAR, visual, and IMU sensors. We use pose information from implementing a keyframe-based SLAM algorithm on forestry data to globally register semantically labeled point clouds. Since the pose updates from the keyframe-based SLAM system are sparse compared to the semantically labeled point clouds, we implement a relative frame pose correction interpolation method that uses keyframe poses as constraints to derive corrected relative frame poses for registration. We demonstrate our approach to real-world data collected using our sensing payload on a UAV operating in forest environments.