This project focuses on creating a versatile path planning system for unmanned aerial vehicles (UAVs) used in outdoor exploration. The system generates a 3D navigation plan based on detected objectives and wind patterns during flight. The approach involves developing a reinforcement learning (RL) policy in a simulation, training it on real aerial imagery, and deploying it on a quadcopter for real-world target detection. This project has future applications in Farming, Search & Rescue, and National Intelligence.
Path Planning Drone with Reinforced Learning
My Contribution
I worked on researching different RL techniques to be used when training and deciding on the type of AL we would like to make. Then I worked with the team on implementing and training it.
Why I picked this Project
I picked this project because I wanted to work with embedded systems and machine learning so I thought this project would be a great combination of the two.
Skills & Knowledge Gained
Technical skills I improved/learned are:
- Git and Github
- Python
- Machine Learning (Reinforcement Learning)