Low Cost Autonomous Amphibious Bird Chasing Robot

Abstract

The use of artificial intelligence and machine learning to create autonomous robot platforms has been spreading into many applications recently, including animal behavior modification. Following this trend, we propose a low-cost, autonomous, and amphibious vehicle to modify the behavior of birds, such as Canadian geese, in commercial areas. Our robot patrols a predefined area set by GPS via an in-house developed Graphical User Interface (GUI). As it patrols this area along a predefined path, a Convolutional Neural Network (CNN) runs a goose detection algorithm to identify geese within a 5 m range. The robot also has basic collision avoidance through a combination of time-of-flight distance sensors and a bumper that detects physical collisions. Our solution is to chase the Canadian geese away from commercial areas frequented by humans such as golf courses before they nest and become territorial. This solution ensures that geese find safer and less disruptive nesting sites in a way that does not harm them. Moreover, the robot collects both locational and behavioral information by taking pictures, which provides information for bird behavior research. Our platform shows the potential to resolve human-animal contested environments with a low-cost intelligent robot solution that can be extended to many other applications.