Embedded Files
Summary
Summary
- Bycatch is the incidental capture of non-target animals in fisheries.
- Bycatch decimates marine species & hurts commercial fishing operations.
- Bycatch Reduction Technologies (BRTs) employ sensory stimuli to deter megafauna from being targeted by commercial fishing gear.
- Current BRTs are not energy efficient and produce waste which further damages the environment.
- Smart Nets are our proposed solution to understand the effects of BRTs, and optimize the implementation of BRTs.
- Smart Nets incorporate machine learning, image recognition and sensory stimuli outputs to provide real-time assessment of BRTs and how they interact with sea life.
Impact
Impact
Scientific
Scientific
- Modelling Highly Dynamic Marine Environments
- Cataloging Behavior Response in Sea Life
- Developing a Autonomous, Multimodal, Closed-Loop CPS
- Identifying Power Efficient Design Parameters for Bycatch Reduction Technologies
Education
Education
- Interdisciplinary Research
- Conservation Biologists and Electrical Engineers at Arizona State University work together to distill design requirements and implement real-world solutions.
- Cross-Community Collaboration
- Trained local communities on light-based BRT s.
- Taught principles of sustainable and renewable energy engineering.
Community
Community
- Field Research
- Worked with regional NGO, Grupo Tortugeuro, to deploy a light-based BRT net with commercial fisheries.
- Support of Local Communities
- Worked with local fisheries to determine commercially viable conservation strategies.
Methods
Methods
From Design Requirements to Real-World Prototype
From Design Requirements to Real-World Prototype
- Distilled Design Requirements from Real-World Fishing Operations
- Built 40 Prototype Light-Based Solar Powered BRTs
- Deployed During Commercial Fishing Season
Programmable Flash
Programmable Flash
- Programmable Duty Cycle For Flashing LEDs via IR Remote
- Allows For Multiple Field Trials with Different Power Profiles
- Lasts ~4 Days While Blinking w/o Charging (15% DC, 1s Period)
- Charges ~30mA/h w/ Adequate Sun Coverage
Species Detection Algorithm
Species Detection Algorithm
Pre-trained using ~25000 images. Currently achieves a 97.2% accuracy in recognition.
Turtle1.aviObject Recognition Example
Object Recognition Example
- Trained to recognize turtles for now.
- Uses a predefined number of frames to provide a confidence level in detection.
- Tunable recognition system to optimize "effort".
IMG_5797.MOVFlashing Buoys
Flashing Buoys
- Each unit flashes at a specified duty cycle.
- Defaults to 15% Duty Cycle w/ a 1s period.
- Programmable via an IR Remote
Page updated
Report abuse