Nappa: "Vegeta! What does the GPU say about the model's power level?" Vegeta: "It's over nine thousand!"
A research simulator testing how small vision-language models control a drone in hazardous 3D environments using camera frames, telemetry, and high-level actions.