Information Theoretic Source Seeking Strategies for Multiagent Plume Tracking in Turbulent Fields

Abstract

We present information theoretic search strategies for single and multi-robot teams to localize the source of biochemical contaminants in turbulent flows. The robots synthesize the information provided by sporadic and intermittent sensor readings to optimize their exploration strategy. By leveraging the spatio-temporal sensing capabilities of a mobile sensing network, our strategies result in control actions that maximize the information gained by the team while optimizing the time spent localizing the position of the biochemical source. By leveraging the team's ability to obtain simultaneous measurements at different locations, we show how a multi-robot team is able to speed up the search process resulting in a collaborative information theoretic search strategy. We validate our proposed strategies in both simulations and experiments.

Exploiting Stochasticity for the Control of Transitions in Gyre Flows

“…(8) When c = 0, the action R 0 is given by the uncontrolled case (4), and the corresponding escape path is x 0 (t). Note that, for 1 Due to the nature of the noise driven transition t 0 = −∞ and t f = ∞.…”

“…Examples include characterizing the dynamics of plankton assemblages [1], measurement of temperature profiles [2], and monitoring of harmful algae blooms [3] and the dispersion of harmful contaminants [4]. These applications demand persistent monitoring of the relevant processes, and as such require the AMV-based sensors to operate for long durations of time.…”

Decentralized Multi-agent information-theoretic control for target estimation and localization: finding gas leaks

“…State-of-the-art methods for information-theoretic-based target estimation and localization primarily focus on using a single agent or assume a central planner to coordinate multiple robots, e.g., see the work by Hoffmann and Tomlin (2010), Charrow et al (2014a, b), Cliff et al (2018), Bayat et al (2016), Hajieghrary et al (2017), Ryan and Hedrick (2010), Bourne and Leang (2017); Bourne et al (2019), Ristic et al (2017), and Park and Oh (2020). Centralized systems are often used even though practical real-world implementation requires decentralized estimation, planning, control, and collision avoidance (Durrant-Whyte, 2006).…”

“…In a centralized system, a single belief distribution is only updated (Bayat et al, 2016; Bourne and Leang, 2017; Bourne et al, 2019; Charrow et al, 2014b; Hajieghrary et al, 2017; Hoffmann and Tomlin, 2010; Hutchinson et al, 2018; Park and Oh, 2020; Ristic et al, 2017; Vergassola et al, 2007). The majority of existing methods also plan in a myopic sense (i.e., planning only for one time-step ahead) (Bayat et al, 2016; Bourne and Leang, 2017; Bourne et al, 2019; Hajieghrary et al, 2017; Hoffmann and Tomlin, 2010; Hutchinson et al, 2018; Park and Oh, 2020; Ristic et al, 2017; Vergassola et al, 2007). In addition, prior works make simplifying assumptions (Bayat et al, 2016; Bourne and Leang, 2017; Hajieghrary et al, 2017), for instance assuming a single-mode belief distribution (Atanasov et al, 2015; Choi and How, 2010; Grocholsky et al, 2003).…”

“…The majority of existing methods also plan in a myopic sense (i.e., planning only for one time-step ahead) (Bayat et al, 2016; Bourne and Leang, 2017; Bourne et al, 2019; Hajieghrary et al, 2017; Hoffmann and Tomlin, 2010; Hutchinson et al, 2018; Park and Oh, 2020; Ristic et al, 2017; Vergassola et al, 2007). In addition, prior works make simplifying assumptions (Bayat et al, 2016; Bourne and Leang, 2017; Hajieghrary et al, 2017), for instance assuming a single-mode belief distribution (Atanasov et al, 2015; Choi and How, 2010; Grocholsky et al, 2003). In such cases, linearization and Kalman filtering techniques are often used even though nonlinearities and multi-modal probability distributions are common in target-tracking problems (Atanasov et al, 2015; Choi and How, 2010; Grocholsky et al, 2003).…”