Our build and extension of the Stanford Pupper Robot. Codename: C.E.R.B.A.R.I.S.
Introduction | Team | Hardware | Software Overview | Software Setup | Computer Vision | Collision Avoidance | Web Interface | Odometry | Behavioral Control |
As a team we set out to build the Standford Pupper robot and see how far we could extend the software, hardware and overall functionality of the system over the course of a semester.
We initally set out to create a robot that could freely move around an enviroment autonomously and seek out visual targets without colliding with any objects. While we accomplished this goal, we had also hoped, but were unable, to fully implement odometry and environment mapping. So though Cerbaris can autonomously seek out a designated target, it cannot – yet – do so in an efficient and robust manner.
Our team upgraded the Pupper by integrating a front-facing camera, a 9-DOF inertial measurement unit (IMU), and 3 digital IR proximity sensors. With the aid of a Google Coral TPU, camera data was utilized to quickly and reliably identify human and tennis ball targets in real-time. We then created software to have the pupper randomly search a room while avoiding obstacles until it finds and chases down the target. Additionally, we were able to create a web-based real-time monitor of Cerbaris’s internal state.