Python vs Orangutan
summary
Six animals need locating across 2,000 km² so a drone-mounted receiver system is built that uses a software-defined radio feeding an Odroid running Python (DroneKit, numpy, Flask) and detection by FFT and cross-correlation. Implanted VHF tags pulse at unique frequencies with ~200–300 m ground range; tests show ~100 m in UK and ~50–60 m in Borneo. Flights run 2 km out on lawnmower patterns, produce heatmaps and successfully locate animals, but hardware failures and canopy attenuation limit reliability.
AI-written summary · about
chapters (12)
- 0:00Python vs Orangutan project
- 2:12Orangutan conservation challenges
- 6:44Tracking implants limitations
- 10:12Drones as practical solution
- 11:55Drone tracking stack
- 14:21Signal detection in Python
- 18:17Proof of concept tests
- 20:08Field testing in Borneo
- 23:36Hardware failures in field
- 26:04Lessons learned and future
- 27:49Conservation reflections
- 32:35Q&A drone design
related talks
36mKeynote: Python vs Orangutan
31mJaguars and Snakes
23mA Drone Flew Over the Python's Nest
33mFlying by the seat of our pants - our journey of teaching python using drones
15mThe Internet of Wild Things: Acoustics, AI and Conservation
25mTechnology gone wild
31mBuilding a (semi) Autonomous Drone with Python
10mDeep Learning in the Wild: Embedded PyTorch for...