I.J. Intelligent Systems and Applications, 2013, 09, 47-57 Published Online August 2013 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijisa.2013.09.06 Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 09, 47-57 Formulation of FISPLAN: A Fuzzy Logic based Reactive Planner for AUVs towards Situation Aware Control Subhra Kanti Das Robotics & Automation, CSIR-CMERI, M G Avenue, Durgapur, WB, INDIA, PIN-713209 E-mail: subhrakanti.das82@gmail.com Dibyendu Pal Robotics & Automation, CSIR-CMERI, M G Avenue, Durgapur, WB, INDIA, PIN-713209 E-mail: subhrakanti.das82@gmail.com Abstract — the paper presents a detailed discussion on the structural organisation of a Fuzzy Inference System Planner (FISPLAN) for Autonomous Underwater Vehicles (AUVs), including elaboration of membership functions for the inputs as well as outputs. The inference mechanism is detailed with discussions on the rule base, which in essence incorporates the planning logic. In order to assess the effectiveness of the planner as a means of reactive escape under critical situations, a case study is studied with reference to a state of the art AUV. An approximate subsea current model is developed from field observations, and residual energy is estimated by referring to a typical Lithium-polymer cell discharge characteristic together with data recorded in actual field trials. Situations are simulated by considering different combinations of sea-currents as well as status of resident energy. Results reveal that the simulated system, by virtue of the planner, is capable of perceiving situations, thereby realizing their imminence and making a decisive action thereupon. In concise, the fuzzy planner may be considered to provide human-like perception of situations on the basis of crisp observations. Furthermore dynamics of the system are modelled with actual parameters, and subsequently controller responses for pitching and velocity correction are illustrated. Choice of planning interval is also expressed as a function of the controllers’ response. Index Terms— Planning, Reactive Architecture, Fuzzy, Situation Awareness, Escape I. Introduction Over the past few decades, an increasing demand of oceanographic explorations have given rise to the development of technologies particularly suited for unmanned underwater operations, out of which AUVs have framed out a niche as the most cost-effective and sustainable technology over the coming years. AUVs stand for autonomous underwater vehicles. These are mobile robotic systems, which can operate underwater at great depths (ranging from 150 to 6000 meters) with limited human intervention or supervisory control [1]. The criticality of operation of such self-controlled systems lies in the design of robust software architecture, involving both planning and control existing in the form of a set of well-coordinated functional modules interfacing with various sensors, actuators and associated controllers present in the system. Although for most practical cases, control and planning modules lie diffused existing as a single entity, it is significant enough to go for both structural and functional decomposition of the processes into mutually cooperative and individually identifiable components. Originating from this underlying concept several diversified architectures have been proposed and worked out in the literature. However, most of them can be categorized into (1) reactive [2]-[5] (2) deliberative [6]-[8] as well as the recently formulated (3) hybrid deliberative and reactive [9], [10], [11], [12]. Contrastingly, few works [13], [14] have incorporated probabilistic planning in order to achieve the mission goals in a completely non-deterministic environment. The proposed approach centers on the formulation of the navigation problem as Partially Observable Markov Decision Problem (POMDP). This inherently reinforces the control framework towards dynamic uncertainties of the environment. However, as a still more recent approach, a new concept of situation awareness with semantic representation, has been introduced [15] with the idea that heterogeneous real-world data of very different type must be processed by and run through several different layers, to be finally available in a suited format and at the right place to be accessible by high-level decision-making agents. The rest of the document is organized as follows-the subsequent section presents a detailed discussion on the structural organization of the Fuzzy planner. Membership functions are elaborated and the rule base is dissected for enforcement of the planning logic. The