Marine Engineering Frontiers (MEF) Volume 1 Issue 1, February 2013 www.seipub.org/mef 1 Stability Augmentation and Fault Tolerance for a Hexapod Underwater Vehicle Olivia Chiu *1 , Meyer Nahon *1 , Nicolas Plamondon 3 Centre for Intelligent Machines, McGill University 3480 University St., Montreal, Canada, H3A 2A7. 1 omychiu@gmail.com; *2 Meyer.Nahon@mcgill.ca; 3 Nicolas.Plamondon@pwc.ca Abstract AQUA is an underwater hexapod robot that uses paddles to propel and orient itself. The system is typically operated remotely by a pilot, with feedback from cameras and on- board sensors. In this work, a stability augmentation system was developed and evaluated on the robot. In order to study the stability of the system, its model was linearized about a nominal equilibrium at several different cruising speeds. Since the robot is never truly in equilibrium due to its oscillating paddles, this required a novel approach. The stability of the unaugmented vehicle was evaluated and improved using sensor feedback. The stability augmentation system was then modified to compensate for possible faults that could occur during the operation of the robot. The failure of a leg was investigated by analyzing the additional drag forces created by the fault. The controller was implemented on the robot with encouraging results. Keywords Underwater Vehicles; Biomimetic; Fault Tolerant Control; Stability Augmentation. Introduction AQUA is a six-legged amphibious robot, shown in FIG. 1, which can swim with the use of oscillating paddles. With these flippers, the robot is able to directly control roll, pitch, yaw, surge and heave. This makes AQUA unique compared to other underwater vehicles which use thrusters to propel themselves. FIG. 1 THE AQUA HEXAPOD ROBOT The environment in which AQUA operates is often unpredictable and can perturb the robot. One way to reduce the influence of the external forces on the system and thus stabilizing it, is through the use of a stability augmentation system (SAS). This type of system is widely used in flight control to aid pilots and to improve the response of a vehicle (Fullmer et al., 1992; Oliva, 1994; Kahn, 2003). Such systems are not as commonly found in underwater vehicles and the issues of stability are often incorporated into the design of the tracker rather creating a separate controller (Nakamura and Savant, 1992; Do et al., 2004; Licht et al., 2007). Previous work on trajectory tracking controllers for ‘conventional’ underwater vehicles is extensive. A survey was done by Yuh (2000) on the design and control of Autonomous Underwater Vehicles (AUVs). These included sliding, nonlinear, adaptive, neural network and fuzzy control. Other experimental comparative studies have also been done by Lea et al. (1999) and Smallwood et al. (2004). They implemented different types of controllers on an AUV or ROV (Remotely Operated Vehicle) and compared the performance and complexity of each controller. This is by no means a complete list of all controllers available for underwater vehicles, however it must be emphasized that all the above controllers were designed for systems that utilized thrusters for propulsion. There have been very few studies into controlling foil-based vehicles, which include work by Hsu et al. (2003) and Licht et al. (2007). Prior work on fault tolerance falls into two main categories: fault detection or control reconfiguration. Research on fault detection usually consists of comparing the behavior of the vehicle to a model and identifying discrepancies. A fault is detected when the discrepancies exceed a predetermined threshold (Orrick et al., 1994; Rae and Dunn, 1994). A common approach to compensating for faults is to design robust controllers (Fossen and Balchen, 1988; Leonard,