Autonomous drone simulator that maps environments and obeys English commands
It provides a full‑stack quadrotor simulation that builds a 3‑D occupancy map from an onboard RGB‑D camera, runs a real cascaded flight controller at 240 Hz, and accepts plain‑English directives via a local LLM. The system runs entirely on a desktop—no cloud APIs or datasets—making it ideal for researchers and hobbyists experimenting with vision‑based autonomy. By integrating perception, planning, and control in one open‑source package, it saves users the effort of stitching together disparate tools. Its self‑contained design and reproducible results set it apart from fragmented demos.
View on GitHub →MichaelFowler1/autonomous-drone-sim