A Computational System to Support Fully Automated Mark-Recapture Studies of Ants

Ants are ubiquitous insects that have great significant for humans. In agriculture they can suppress pest populations and aerate the soil, but they sometimes protect pests, leading to billions of dollars of crop loss. As such, there is a growing need to study these insects, both in the lab and in the wild. In this work we describe an end-to-end machine learning/robotic system to physically capture individually marked insects. To the best of our knowledge, this is the first system described in the literature that can capture individually targeted insects, without harming them, allowing them to be recaptured multiple times.