Overview
This lecture introduced the development and testing of an animal-borne acoustic shockwave detector designed to combat wildlife poaching by detecting gunshots and providing real-time alerts. The talk also covered a related low-power acoustic data logger, AI methods for audio filtering, and addressed practical aspects of deployment and cost.
Motivation & Background
- Poaching is a major issue threatening wildlife, particularly through gunshot hunting.
- Existing static gunshot detector systems are limited in coverage and do not directly protect animals.
- Advances in GPS collar technology allow for animal-borne sensor integration.
Gunshot Detection Technology
- Gunshots produce both a muzzle blast and a supersonic shockwave (sonic boom).
- Shockwave signals are unique and easier to detect than other acoustic events.
- Multiple microphones in an array can determine shooter direction, bullet trajectory, and other details, but the animal-borne sensor uses a single microphone for shockwave detection only.
System Design & Prototyping
- The shot detector is designed as an add-on to commercial GPS collars.
- Upon detecting a shockwave, the device immediately sends an alarm with the latest GPS data.
- Hardware uses both a regular microphone and a piezo (PSO) sensor to distinguish between real shockwaves and mechanical impacts.
- Includes an accelerometer to monitor behavioral changes after a detected shot.
Field Testing & Results
- Prototypes tested on cattle in Tennessee, elephants at the San Diego Zoo, and in Kenya.
- System showed high accuracy in distinguishing gunshots from animal and environmental noises.
- Current detection radius is about 50 meters, potentially protecting entire herds.
Ongoing & Related Projects
- A new, smaller, low-power acoustic data logger is being developed.
- Features include scheduled, threshold, and AI-based event-triggered recording.
- AI uses unsupervised clustering to group similar sounds, reducing storage of redundant audio.
- Target is year-long operation depending on battery and usage settings.
Q&A & Practical Considerations
- AI model uses unsupervised clustering, not traditional classification, with user-configurable parameters.
- Device cost expected to be under $100 in medium-scale production.
- Animal-borne detector uses only a single microphone and PSO sensor—does not provide shooter location.
Key Terms & Definitions
- Shockwave — A rapid pressure wave produced by a supersonic bullet, detected as an acoustic signature.
- Muzzle blast — The explosive sound from the gun’s barrel immediately after firing.
- Piezo (PSO) sensor — A sensor detecting mechanical vibrations, used here to differentiate shockwaves from impacts.
- Unsupervised clustering — An AI method for grouping similar data (e.g., sounds) without predefined categories.
Action Items / Next Steps
- Finalize new hardware design and update detection algorithms with AI techniques.
- Integrate shockwave detector into commercial collars (e.g., with Sava Tracking, Kenya).
- Continue field testing and open-source release of hardware and software designs.