Jason Ye
Research Mentor: Kimberlee Kearfott
Mentor Department: NERS, Engineering
Author(s): Jason Ye, Adam Drihany, Kimberlee Kearfott
Session: Session 7 (4:00 PM – 4:50 PM)
Presentation Type: Poster 10
Abstract
Autonomous drones can rapidly characterize radiation fields while reducing worker exposure, but safe operation in low-altitude or cluttered environments requires a reliable anti-collision framework. The long-term goal of this project is to develop a fully integrated anti-collision system for an Intelligent Radiation Awareness Drone. This framework can be divided into several phases: validating and calibrating sensor input data before flight, detecting potential collisions from live sensor measurements, enabling autonomous real-time collection and processing of collision-relevant data during flight, and integrating these capabilities with the flight-control system to support avoidance maneuvers. Establishing reliable sensor interpretation and communication is a necessary first step toward this broader autonomous navigation objective. This phase of the project focused on building the software and sensing foundation needed for later anti-collision development. QGroundControl, PX4, Gazebo Sim, and the Pymavlink Python library for MAVLink communication were installed and configured to create a working drone simulation environment. Integration between Gazebo Sim and QGroundControl was established to support simulation-based testing and live telemetry monitoring. PX4 and QGroundControl parameters were then tested in Gazebo Sim to evaluate low-altitude flight modes, including terrain following and terrain hold. Terrain following was selected over terrain hold because it better maintains a constant height above ground while the vehicle moves over changing terrain, whereas terrain hold is primarily intended for low-altitude stationary flight. In addition, a Python script using Pymavlink was developed to read live sensor data from the simulated drone in Gazebo simulation, providing a workable framework for sensor validation and telemetry analysis. A LiDAR sensor was also installed and mounted on the drone platform to support future sensor-comparison and navigation experiments. Future work will extend this platform into a complete anti-collision system by developing an algorithm that uses live sensor data to detect and respond to potential obstacles during flight. This algorithm will be optimized for both safe path selection and efficient route planning, while also reducing the delay between sensor input, decision-making, and flight response. Additional work will apply the telemetry and sensing framework to physical drone testing, including comparison of Pixhawk barometer and LiDAR sensor data to assess which sensing approach provides more reliable altitude information during low-altitude operation. These efforts will support the eventual development of autonomous terrain-aware navigation and collision-avoidance behaviors for radiation-survey drone missions.
