2019/07/15 by Minghan Wei, Volkan Isler, Wei, Minghan +1 · 2 citations
Environmental Science · Engineering · Economics, Econometrics and Finance · #Species Distribution and Climate Change #Robotics and Sensor-Based Localization #Economic and Environmental Valuation
paper · pdf · doi:10.48550/arxiv.1907.06337
As mobile robots find increasing use in outdoor applications, designing\nenergy-efficient robot navigation algorithms is gaining importance. There are\ntwo primary approaches to energy efficient navigation: Offline approaches rely\non a previously built energy map as input to a path planner. Obtaining energy\nmaps for large environments is challenging. Alternatively, the robot can\nnavigate in an online fashion and build the map as it navigates. Online\nnavigation in unknown environments with only local information is still a\nchallenging research problem. In this paper, we present a novel approach which\naddresses both of these challenges. Our approach starts with a segmented aerial\nimage of the environment. We show that a coarse energy map can be built from\nthe segmentation. However, the absolute energy value for a specific terrain\ntype (e.g. grass) can vary across environments. Therefore, rather than using\nthis energy map directly, we use it to build the covariance function for a\nGaussian Process (GP) based representation of the environment. In the online\nphase, energy measurements collected during navigation are used for estimating\nenergy profiles across the environment using GP regression. Coupled with an\nA*-like navigation algorithm, we show in simulations that our approach\noutperforms representative baseline approaches. We also present results from\nfield experiments which demonstrate the practical applicability of our method.\n