Autonomous Drone for Nuclear Power Plant Inspection

Autonomous Drone system designed for our final Capstone Project

The Drone autonomously flying to different waypoints

For our final year capstone project, my team and I build and programmed an autonomous drone system for a Nuclear Power Plant inspection task. The drone was flown in a drone lab at the University in a room full of obstacles which were not known apriori.

For this project, our team of 4 split ourselves such that I worked on the path planning, obstacle avoidance, and was in charge of the testing tuning of our drone and software parameters. I also was responsible of the adminstrative aspect.

The project was split up into 5 deliverables, starting from autonomous takeoff and landing, hovering on the spot, following a simple trajectory with no obstacles and finally autonomously navigating the environment to reach waypoints.

We used a realsense T265 for our visual odometry, a pixhawk flight controller and MAVROS, a library that enables us to receive telemetry from the pixhawk onto our ROS network. We used an onboard Jetson Nano for our computation. A lidar sensor was also added for altitude estimation. A color camera was used to perform obstacle detection.

An annotated diagram of our drone.
Some pictures of the view of the drone's obstacle avoidance system in action, the obstacle-free flight trajectory and the custom camera mount modification to angle the camera to perform obstacle avoidance.