Robotic Fish: Autonomous Navigation

A robotic fish swims freely in an outdoor pool without human intervention or control. A computer vision system tracks the fish, and frequently broadcasts data to the fish using a custom low-frequency radio. This system is necessary because GPS signals do not propagate through water and because the fish has no means to sense the walls. The vision system estimates the x-y-z position, 3D orientation, and speed of the fish using a Bayesian particle filter whose measurements come from four calibrated underwater cameras. Adaptive image processing was needed to achieve consistent results.

Category:

Education

Tags:

• robot
• robotic fish
• vision system
• particle filter