verily | Debug Trap Imaging

Context

Verily's Debug program uses mosquito traps to test the impact of their population control. Monitoring and classification of the trapped bugs requires significant time of well-trained entymologists.

Goal

Demonstrate proof-of-concept for low-cost imaging hardware to speciate and gender mosquitos caught in traps

My role

PD Lead

Sensor

Off-the-shelf, low-cost imaging sensor

Sensor function

Identify mosquito species and gender

My contributions

• Identify sensor requirements (resolution, field of view, MTF, depth of field) and acquire low-cost hardware

• Program Arduino microcontroller and build prototypes

• Wrote MATLAB scripts that enabled tuning parameters for noise, speed, and image quality

• Generated 3D CAD of the trap, imaging and lighting hardware, enclosure and attachment mechanism

Prototypes

• Simple, bench-top, computer-connected with enclosure

Fabrication Techniques

• Flex assembly

• SLA

• Threaded fasteners

Outcomes

• Identified multiple system options that achieved the necessary image quality in 3 month timespan

• Software teammates were able to begin training AI-powered algorithms to classify and gender mosquitoes from the imaging system.