2016 Vehicle: Dory

"Dory" is an Autonomous Underwater Vehicle (AUV) designed to compete in the 19th annual AUVSI/ONR RoboSub Competition . Dory is the Caltech Robotics Team’s most advanced and capable vehicle, equipped with a gimbaled camera, seven thrusters, two gripper mechanisms, a hydrophone array, and a sophisticated set of sensors.  Dory represents the culmination of the team’s efforts to design, build, test, and refine a customized system, providing students with valuable industry skills and technical expertise.

Main Assembly Colored Resized

Mechanical System

Dory's mechanical design features a single pressure hull which houses all of the onboard electronic and pneumatic components. This pressure hull has a curved acrylic dome that seals with a 3-dimensional o-ring, allowing clear visibility of the interior components.  Attached to the pressure hull are seven VideoRay M5 Brushless thrusters that provide control over all six spatial degrees of freedom with extra redundancy.  A Guppy Pro F-046 camera is housed within a custom gimbal system, which improves object detection and allows for real-time tracking.  A pressurized pneumatic system allows the vehicle to drop markers, fire torpedoes, and actuate both two degree-of-freedom grippers.   (more)

Electrical System

Dory’s sensors include (1) A VectorNav VN-100 Attitude Heading and Reference System, (2) A SonTek Argonaut Doppler Velocity Log, (3) Four Teledyne Reson TC4013 Hydrophones, and (4) An Omega PX309 Pressure Transducer.  Two 6S LiPo batteries separately power the vehicle’s actuation and computation circuits.  An Intel NUC with a Core i3 processor communicates with five custom PCBs that collect sensor and actuator data over a CAN bus network.  (more)

Software System

Dory’s software runs on an ubuntu environment, using a custom architecture consisting of Hardware, Mobility, Vision, Commander, and Logger processes.  Information is passed between these processes using a publisher-subscriber model.  The hardware process interprets data from the sensors and actuators, then publishes the data to the robot state topic.  The mobility controller uses a linear quadratic regulator model to provide excellent stability in the water. After images from the gimbaled camera are adjusted for exposure, white balance, and distortion, a rigorous feature recognition algorithm is applied to identify key objects.  (more)

  

Main Assembly Colored Open

Login Form

Joomla templates by Joomlashine