Abstract
Forestry is one of the most hazardous occupations in the United States, with forest operations having high rates of workplace injuries and fatalities. Emerging SMART technologies such as artificial intelligence (AI), cloud computing, advanced spatial data processing, and robotic vision offer new opportunities which could enhance operational safety and precision forestry. This thesis examines how some of these technologies can be utilized in operational forestry. Chapter 2 evaluates the use of mounted stereo cameras with video AI to identify ground personnel when they walk through the cameras field of view. In field experiments analyze how basal area (stand density), distance, and treatment conditions influence how effective identifying ground personnel is. Chapter 3 explores an alternative approach to positioning that utilizes terrestrial laser scanning (TLS) and airborne laser scanning (ALS) derived tree positions to estimate location of forest resources. Often, GNSS systems are effected by multipath error and canopy interceptions causing high horizontal error when projecting a ground position. The alternative approach would be using segmented predicted tree locations from the ALS and TLS datasets and applying point matching techniques to predict forest resource position. This work proposes possible solutions to increase occupational safety on forest logging operations through video AI on stereo cameras. Additionally, it offers potential alternatives to GNSS systems by using predicted tree location map-matching from lidar acquisitions. This work contributes to a greater movement of integrating SMART devices in forest operations.