Fig. 1. Modified OceanServer IVER2 AUV Abstract— A critical problem in planning sampling paths for autonomous underwater vehicles is balancing obtaining an accurate scalar field estimation against efficiently utilizing the stored energy capacity of the sampling vehicle. Adaptive sampling approaches can only provide solutions when real-time and a priori environmental data is available. Through utilizing a cost-evaluation function to experimentally evaluate various sampling path strategies for a wide range of scalar fields and sampling densities, it is found that a systematic spiral sampling path strategy is optimal for high-variance scalar fields for all sampling densities and low-variance scalar fields when sampling is sparse. The random spiral sampling path strategy is found to be optimal for low-variance scalar fields when sampling is dense. I. INTRODUCTION utonomous underwater vehicles (AUV) are mobile robotic platforms that are utilized for environmental sampling, sensing and are capable of operating in a multitude of underwater aquatic environments[1]. In research they are commonly used to carry hydrological, geophysical, and/or biological sensor payloads that are used to gather remote and in-situ data to aid the study of open and closed aquatic bodies [2]. An autonomous underwater vehicle is generally described as being capable of operation without any human controller; thus it must be able to traverse the aquatic environment on a desired path through autonomous navigation and control [3]. An example of an autonomous underwater vehicle is the one used in this paper to collect the data that was then processed to generate scalar fields (see Fig. 1). The goal of environmental sampling and sensing is to generate an accurate estimate of the underlying scalar field(s) over an area of interest. An estimate is generated by interpolating the sensing data collected over the area being studied. In the case of aquatic environments, a mobile vehicle can be used to collect the necessary sensing data. However the sensing platform is limited in range by the amount of stored energy it carries aboard. The problem of environmental sensing then becomes determining how one obtains an accurate estimation of the scalar field of interest while being constrained by the amount of stored energy the vehicle can utilize in order to sample the scalar field. The purpose of this paper is to experimentally evaluate sampling strategies based upon estimation accuracy and energy consumption. This evaluation is conducted with a cost-evaluation function that considers multiple parameters, Colin Ho, Andres Mora and Srikanth Saripalli are with the School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85284 USA (e-mail: colinho@asu.edu). and can assign priorities to each factor. The evaluation takes into account real world and simulated isotropic, anisotropic scalar fields, as well as varying sampling densities to determine which sampling strategy is optimal for different scalar field types, for a range of sampling densities. The sampling path strategies evaluated in this study were systematic and stratified random sampling distributions with spiral and lawn mower sampling paths. Through utilizing the cost-evaluation function, it is found that the systematic spiral sampling path strategy best optimizes the energy consumption with the scalar reconstruction error compared to the other sampling path strategies evaluated. Additionally, it is found that the systematic spiral path consumes the least amount of energy out of the sampling path strategies evaluated. II. RELATED WORK In the fields of robotics, hydrology, geology, and geostatistical sciences, optimal sample collection and path planning are an active area of research [4]. There is much prior and current work ongoing in the domain of path planning and sampling optimization for autonomous vehicles. For example, adaptive sampling algorithms have been developed that can direct the path of single or multiple autonomous underwater vehicles in towards locations of high probable data yield [5], and can be used in conjunction with existing sensor networks [6]. Additionally, energy optimal paths can be computed based upon known and sensed external variables such as ocean currents [7] [8], and static or dynamic obstacles [9]. An Experimental Evaluation of Various Sampling Path Strategies for an Autonomous Underwater Vehicle Colin Ho, Andres Mora and Srikanth Saripalli A