“…Information-theoretic control (or active perception) is a technique to move a sensor so that more informative observations are obtained to improve Bayesian estimation of an unknown target of interest (Julian et al, 2012). This technique has been used in a wide range of applications, e.g., selecting favorable placement of static sensors (Cameron and Durrant-Whyte, 1990), simultaneous localization and mapping (SLAM) in GPS-denied environments (Atanasov et al, 2015), estimating agent states in a multi-agent system (Hausman et al, 2015), finding/tracking unknown number of stationary and mobile targets (Dames and Kumar, 2015; Dames et al, 2017; Grocholsky et al, 2003; Ryan and Hedrick, 2010), generating 3D maps of an unknown environment (e.g., in a cave) (Charrow et al, 2015a, b ; Tabib et al, 2016), inferring the state of a forest fire (Julian et al, 2012), tracking animal migration patterns (Cliff et al, 2018), finding victims buried in a snow avalanche (Bourne and Leang, 2019; Hoffmann and Tomlin, 2010), locating magnetic anomalies such as mineral deposits and bomb ordnance (Dames et al, 2016), tracking invasive species such as algae, carp, and weeds (Clements et al, 2014; Dunbabin and Marques, 2012; Tokekar et al, 2013), estimating mass properties during cooperative lifting tasks (Corah and Michael, 2017), maintaining object visibility for visual servoing (Dame and Marchand, 2011), modeling objects, e.g., for manipulator grasping (Denzler and Brown, 2002; Kahn et al, 2015; Whaite and Ferrie, 1997), forecasting the weather (Choi and How, 2010), and localizing a chemical (Bayat et al, 2017; Bourne and Leang, 2017; Hajieghrary et al, 2017; Neumann et al, 2013; Ristic et al, 2017), radioactive (Ristic et al, 2010), radio (Charrow et al, 2014b), or acoustic source (Basiri et al, 2018).…”