Robotic Tuna Research
Fall 2014
There were two goals of research on the robotic tuna during the fall semester:
1. Transitioning to Computation on Board
To transition the fish to having computation on board, we chose to use the NI myRIO, a control board that includes both an FPGA layer as well as a real-time layer. Both layers are programmed through LabVIEW.
The brain structure for the robotic fish is an adaptation of the Robot-In-A-Box brain architecture that is used by numerous projects in the Olin Intelligent Vehicles Laboratory at Olin College. In essence, this brain architecture is the result of combining the Sense-Think-Act paradigm as well as the Hindbrain-Midbrain-Forebrain structure of human brains.
The following is the current version of the brain for the robotic fish:
2. Developing the Structure Light System for Obstacle Detection
Because traditional methods of obstacle detection (e.g. infrared sensors) no longer work under water, we needed to develop a new way to detect obstacles under water as well as seek out walls to implement a wall-following behavior. Thus, we decided to use a detection method known as structured light. In essence, a laser casts a horizontal laser line across the plane of view of the camera. If an object is present, the laser line will be divided into segments of different heights. Using the heights of the line segments and some computation, the distance between the camera and the object can be calculated.
A video of the current version of this system is shown in the video below
Because traditional methods of obstacle detection (e.g. infrared sensors) no longer work under water, we needed to develop a new way to detect obstacles under water as well as seek out walls to implement a wall-following behavior. Thus, we decided to use a detection method known as structured light. In essence, a laser casts a horizontal laser line across the plane of view of the camera. If an object is present, the laser line will be divided into segments of different heights. Using the heights of the line segments and some computation, the distance between the camera and the object can be calculated.
A video of the current version of this system is shown in the video below