An Interactive Robust Artificial Intelligence-based Defense Electro Robot (RAIDER) using a Pan-Tilt-Zoom Camera

The Vision Lab’s Robust Artificial Intelligence-based Defense Electro Robot (RAIDER) is an integrated electro-mechanical system equipped with an onboard processor and numerous imaging sensors. The RAIDER is built upon the Clearpath Husky A200 mobile base. In a multidisciplinary effort, the newly constructed robotic body houses the onboard laptop, GPU processor, LAN, IP cameras, and Kinect sensors. In our previous experiments and efforts, we shown the capability of computing a 3D model of the surrounding scene from motion imagery. We have tested autonomous navigation algorithms in which the RAIDER was to follow a particular person in a crowded environment. Algorithmic enhancements have integrated the 3D depth information into the person-tracking technique to allow for following a person around sharp corners. These navigation and controls algorithms call for an accurate face detection and recognition system as well as a human body detection and recognition system. Additionally, we have integrated a Play Station 2 wireless controller to remotely maneuver the RAIDER and activate various autonomy modes. In this poster, we present our latest effort in integrating face detection with the Pan-Tilt-Zoom (PTZ) base of an Axis camera. Positioned on top of the RAIDER, the PTZ-base will allow for the RAIDER to mimic a human’s ability to “look around” or “follow a person with only the eyes,” specifically without physically turning the robotic body. The face detection algorithm provides the location of a face within the images, the PTZ is constantly tracking the face and adjusting to keep it in the center of the image. Additional RAIDER projects work on integrating a speaker system that would vocalizes pre-defined phrases triggered by the recognition of specific persons. This would allow the RAIDER to vocalize “Hello” to people trained into its recognition system. These new artificial-intelligence RAIDER innovations create a more interactive human-like robotic system.https://ecommons.udayton.edu/stander_posters/1390/thumbnail.jp