Intelligent Interface For Remote Supervision And Control Of Underwater Manipulation

Abstract

Underwater manipulation continues to pose challenges to both robotics and computer technologies. With recent advances in sensor and artificial intelligence technologies, "intelligent" robotics presents a novel and effective solution to coping with problems typically encountered in underwater manipulation tasks. To date, a number of manipulation systems with computer-based self-monitoring and planning capabilities have been demonstrated. While the principal contribution of these systems is to replace, augment, aid or improve operator performance in underwater tasks, overall system performance still depends on the effectiveness of the combination between the robotic system and the human counterpart. To this end, an "intelligent" interface ( <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I^{2}</tex> ) is proposed as a means to dynamically manage information, allocate tasks, and actively mediate between the operator and the robotic manipulator in a manner that optimizes their joint performance. The potential efficacy of the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I^{2}</tex> is discussed in relation to specific aspects of remote supervision in underwater manipulation tasks (e.g., effective data portrayal, plan communication, contingency handling, and task monitoring). An example of a link between AI and a mechanical underwater manipulator is shown via a prototype system, Task-Oriented Supervision Command System (TOSC). TOSC is a procedural net-based supervisory planning system that provides interactive programming of manipulation tasks in a high-level command language. A series of experimental studies were performed to evaluate TOSC features on a remotely-controlled six degree-of-freedom anthropomorphic manipulator system with an integrated control and display console. Laboratory evaluation of underwater manipulation and maintenance tasks at various levels of task complexity and manipulation difficulty was conducted. The results showed TOSC to be a highly effective and well accepted aid that both improved performance and reduced errors, especially in complex manipulation tasks under uncertain environmental conditions.