In this paper, the longitudinal control problem of an underwater vehicle undergoing obstacle avoidance, with large velocity variation, is considered and a gain-scheduled design is presented as a solution. Vehicle dynamics are transformed into a quasi-linear parameter varying form first. Then, the longitudinal control is constructed in two layers. The outer loop plays the role of guidance offering desired velocity command, which is tracked by the inner loop of the control law. The linear parameter varying (LPV) methodology used in the inner loop design circumvents the use of difficult nonlinear control techniques, and cancels the effects of velocity dependent terms. Furthermore, a simplified implementation scheme employing switched LPV control is investigated. Nonlinear simulation demonstrates the effectiveness of the proposed approach.


Underwater vehicles have been intensively studied in last decade. These vehicles can operate in deeper and riskier areas where divers cannot reach. Typical applications are: inspection and structural evaluation of submerged installations, transportation and assembly of underwater structures, ships rescue, mine hunting, geotechnical and environmental data gathering, dumps or toxic waste location, marine archeology and many others (Zanoli and Conte, 2002). The underwater vehicle considered in this paper, named as DELTA, is of shuttle species, the mission of which is to dive to specified depth and measure wide area of the characteristic of water column. In the operational scenario envisioned, the guidance provides a fixed depth, where the change of underwater vehicle dynamics is slow. However, large transients could occur. For example, when the vehicle tries to avoid an obstacle or other swimming species, it has to deviate from its original moving path as soon as possible. In this scenario, the guidance generates an avoidance command. In turn, the vehicle dynamics change suddenly and largely, so a linear time-invariant controller cannot guarantee performance over the entire operating range.

This content is only available via PDF.
You can access this article if you purchase or spend a download.