AI in the Sky

Sea Turtles are found on beaches worldwide. Monitoring of marine turtle nesting and hatching is most often done by biologists and volunteers, who walk the beach at dawn in order to observe sea turtle tracks in the sand, and follow them to nests that were created during the night. Effectiveness of this method depends on the number of beach walkers, level of expertise, and degree of focus. For some nesting beaches, this may be adequate, but for beaches subjected to heavy predation, recreational activity, remoteness, or poaching, significant benefits may be achieved through more timely surveillance.

Small Unmanned Aerial Systems (sUAS), also referred to as drones, offer the potential for better timeliness, consistency, overall effectiveness, and more widespread application. Drones are able to perform surveillance of marine turtle nesting throughout day and night under most weather conditions. KwF has already demonstrated simple drone streaming video to ground operators.

Drones have been around for more than 70 years, but they evolved quickly during that past few years with the introduction of high performance, low power consumption microprocessors originally developed for smart phones and tablets.

Kashmir Robotics, the science and technology division of KwF, is a world leader in the application of embedded systems -- microprocessors distributed throughout the drone -- to achieve a very high degree of autonomy. Drones also have benefited from the development of high performance single board computers. This enables onboard processing of imagery and other data, eliminating the need for streaming large quantities of data to ground computers for post mission processing.

In 2015, Kashmir Robotics began research and development to design, fabricate, integrate, and operate “MiSHELL,” a fully autonomous custom drone with the ability to collect and process data onboard. MiSHELL is being “trained” to perform the functions of human beach walkers by identifying, characterizing, locating turtle tracks, following the tracks to locate nests, observing nests, and report their location and condition, which could be either a nest or a false crawl. Additional pattern recognition and reporting capabilities will be added as required by the marine turtle research and protection community.

Join us to learn about how Malini and other KwF Interns are working on developing A.I. that can detect and identy sea turtle tracks using "You Only Look Once" (YOLO) a state-of-the-art, real-time object detection system. 

Presented By: Malini Shivaram 

Malini Shivaram is an incoming sophomore at Carnegie Mellon University, majoring in Artificial Intelligence. As an AI KwF intern working on image recognition for the MiSHELL sea turtle drone project, she has enjoyed learning about new tools and interacting with other college students on the AI team. Being able to be a part of a project that has an impact on protecting wildlife is an exciting opportunity for her, and she hopes to get more involved in wildlife conservation with AI in the future.