USDA Forest Service selects Chiral Software to develop embedded machine learning system

USDA Forest Service – Savannah River Research Team has selected Chiral Software to develop CoyoteCam, an embedded machine learning system which performs wildlife management tasks. This system, developed under earlier research contracts with USDA Forest Service, will now be deployed to “secure” an area of approximately one square mile to prevent entry of coyotes, while allowing all other animals to enter freely.

The coyote is the apex predator in the ecosystem at the Savannah River location. By excluding just this animal, but letting others enter freely, the Forest Service researchers will observe what happens to an ecosystem without its apex predator. Use of an embedded, autonomous, solar powered machine learning vision system is the only way this could be achieved without causing harm to animals or the ecosystem. The NVidia module is connected to Chiral developed electronics which allow it to control an electric curtain, which is the same as an electric fence but allows animals to pass through easily when it is not active.

CoyoteCam uses Ethernet cameras and NVidia embedded systems, with customized software and training sets, to achieve this goal. The systems are compact, solar powered, and can run indefinitely. Chiral’s team will deploy to the Savannah River Site (SRS) to deploy a set of these systems.

This same technology will be applied to other challenges in wildlife management and defense.