Underwater robots can be used to inspect water-bound structures such as ship hulls. Remote-operating these vehicles can be tricky if they are under-actuated, meaning you can’t fully control their motion. For example, you might command the robot to go forward but uncontrolled sway could result in the robot moving sideways. This can be challenging if you’re trying to inspect an object using a camera, and the object keeps slipping out of your field of view.
To solve this problem, Karras et al. propose a semi-autonomous control scheme that ensures the robot doesn’t lose sight of the inspection target. The control fuses information (using an asynchronous Modified Dual Unscented Kalman Filter) from sensors on the robot to estimate its position and attitude and correct its trajectory when needed.
Experiments were conducted in a test tank using an under-actuated underwater robot that uses only three thrusters. Information for sensor fusion is provided by an Inertial Measurement Unit (IMU), a camera, and two laser pointers which are parallel to the camera axis. By monitoring where the lasers point using the camera, the robot can figure out how it is moving with respect to the inspection target.
Results show the feasibility and applicability of the semi-autonomous control scheme. In the future, the authors hope to extend their approach to more difficult tasks such as inspecting fish farm nets.