Elsevier

Information Fusion

Volume 58, June 2020, Pages 13-23
Information Fusion

Decentralised multi-platform search for a hazardous source in a turbulent flow

https://doi.org/10.1016/j.inffus.2019.12.011Get rights and content

Highlights

  • Autonomous search in a turbulent flow for a source of a toxic release.

  • Multiple platforms connected in a dynamic sensor network operating in a decentralised manner.

  • Source parameter estimation carried out using a Rao-Backwellised particle filter.

  • Platform control based on entropy reduction.

  • Group coordination achieved by exchanging data with neighbours only.

Abstract

The paper presents a cognitive strategy that enables an interconnected group of autonomous vehicles (moving robots) to search and localise a source of hazardous emissions (gas, biochemical particles) in a coordinated manner. Dispersion of the emitted substance is assumed to be affected by turbulence, resulting in the absence of concentration gradients. The key feature of the proposed search strategy is that it can be applied in a completely decentralised manner as long as the communication network of autonomous vehicles forms a connected graph. By decentralised operation we mean that each moving robot performs computations (i.e. source estimation and robot motion control) locally. Coordination is achieved by exchanging the data with the neighbours only, in a manner which does not require global knowledge of the communication network topology.

Introduction

The use of autonomous vehicles for environmental monitoring is of great importance, especially when it involves dangerous missions, such as the search and localisation of toxic gas releases. Some recent surveys containing a plethora of relevant references can be found in [1], [2], [3]. Practical implementation of miniature airborne platforms for gas sensing tasks are explored in [4], [5]. Existing theoretical approaches to the problem of searching and localisation of emitting sources, can be loosely divided into three categories: (i) up-flow motion methods, (ii) concentration gradient-based methods and (iii) information gain-based methods. The first category mimics the behaviour of bacteria and insects in their search for food and mates. For example, upon sensing an odour signal, male moths surge upwind in the direction of the flow, but when the odour information vanishes, they exhibit random cross-wind casting or zigzagging until the plume is reacquired [6]. This class of methods inspired robotic searches by various groups, e.g. [7], [8], [9].

Concentration gradient-based methods, also referred to as chemotaxis, seek for the emitting source by following the positive local gradient of the chemical concentration [10]. These strategies are effective close to the source where the odour plume can be considered as a continuous field. Since the source is at the maximum of the concentration field, sometimes these methods are referred to as the extremum seeking algorithms [11], [12], [13].

Both the up-flow motion methods and the concentration gradient methods are simple in that they require only a limited level of spatial perception [14]. Their limitations manifest in the presence of turbulent flows, due to the absence of concentration gradients, when the plume typically consists of time-varying disconnected patches (the effect known as intermittency). The information gain-based methods, referred to as infotaxis [15], have been developed specifically for searching in turbulent flows. In the absence of a smooth distribution of concentration (e.g. due to turbulence), this strategy directs the searching robot(s) towards the highest information gain. As a theoretically principled approach, where the source-parameter estimation is carried out in the Bayesian framework and the searching platform motion control is based on the information-theoretic principles, the infotaxic (or cognitive) search strategies have attracted a great deal of interest [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27].

This paper focuses on an infotaxic coordinated search by a group of autonomous vehicles (platforms) for an emitting source of hazardous substance dispersed by turbulent flow in open terrain environment. The search platforms are equipped with the appropriate sensors for sequential measuring of: (a) the pollutant concentration and (b) the platform location. Due to the turbulent transport of the emitted substance, the concentration measurements are typically sporadic and fluctuating. The searching platforms form a moving sensor network thus enabling the exchange of data and a cooperative behaviour. The multi-robot infotaxis have already been studied in [16], [22], [24], [28], [29]. However, all mentioned references assumed all-to-all (i.e. fully connected) communication network with a centralised fusion and control of the searching group.

In this paper we propose a fully decentralised infotaxic coordinated search. By decentralised operation we mean that each searching sensor platform performs the computations (i.e. source estimation and platform motion control) locally and independently of other platforms. Group coordination for the sake of achieving the (common) task mission is carried out by exchanging the data only with neighbours, in a manner which does not require global knowledge of the communication network topology. Hence, the proposed approach is scalable in the sense that the complexities for sensing, communication, and computing per sensor platform are independent of the sensor network size. In addition, because all sensor platforms are treated equally (no leader-follower hierarchy), this approach is robust to the failure of any of the searching agents. The only requirement for avoiding the break-up of the searching formation is that the communication graph of the sensor network remains connected at all times. Source parameter estimation is carried out sequentially, and on each platform independently, using a newly developed Rao-Blackwellised particle filter. Platform motion control, in the spirit of infotaxis, is based on entropy-reduction and is also carried out independently on every platform. The mathematical models of concentration measurements and platform motion are similar to those in [28], which developed a search algorithm assuming the centralised architecture and all-to-all communication network. In addition, the source estimation algorithm in [28] was a batch algorithm - which would be impossible to use in a decentralised architecture. At the time of publication of this manuscript, another decentralised infotaxic search was reported in [30].

The paper is organised as follows. Mathematical models are presented in Section 2. The method for decentralised sequential estimation of the source parameters is described in Section 3. Next, decentralised formation control of the robotic platforms is explained in Section 4. Our proposed approach is then evaluated in Section 5 using simulated and experimental data. Finally, conclusions are drawn in Section 6.

Section snippets

Mathematical models

Search can be described as a repetitive cycle of sensing, estimation, decision making for motion control and actuation (the execution of motion control). This section introduces the mathematical models of sensing, vehicle motion and communication, necessary for the autonomy enabling functions of estimation and motion control.

Decentralised sequential estimation

We adopt the measurement dissemination based decentralised architecture [31], where the measurement locations3 and the corresponding measured concentration values, i.e. the measurement trios (xki,yki,zki), are exchanged via the communication network. The protocol is iterative. In the first iteration, platform i broadcasts its trio to its neighbours and receives from them their

Decentralised formation control

In decentralised multi-robot search, each platform autonomously makes a decision at time tk1 about its next control vector (or action) uki. Selection of individual actions will be discussed in Section 4.1. Although the same RBPF code is meant to be executed in parallel on every platform, even if the same dataset D1:k is available to all of them, the particle systems on individual platforms Ski, i=1,,N, may not be the same, because the local pseudo-number generators (used by particle filters)

An illustrative run

In order to illustrate the proposed decentralised multi-robot search, let us consider the following scenario. All physical quantities are in arbitrary units (a.u.). The search domain is a square, whose area is 500 × 500, centered at coordinates (0,0). The true source parameters are X0=166, Y0=147 and Q0=5. The centroid of the multi-robot formation is initially at coordinates x0c=225, y0c=240. The formation is initially in the shape of a regular hexagon, whose side length is 20. There are N=7

Conclusions

In this paper, we have proposed the first decentralised infotaxic search algorithm for a group of autonomous robotic platforms. The algorithm allows the platforms to search and locate a source of hazardous emissions in a coordinated manner without the need for a centralised fusion and control system. More precisely, this distributed coordination is achieved by local exchange of measurement data between neighbouring platforms. Similarly, the movement decisions taken by the platforms were reached

CRediT authorship contribution statement

Branko Ristic: Conceptualization, Methodology, Writing - original draft. Christopher Gilliam: Software, Validation. William Moran: Supervision, Writing - review & editing. Jennifer L. Palmer: Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This research is supported in part by the Defence Science and Technology Group through its Strategic Research Initiative on Trusted Autonomous Systems.

References (43)

  • M. Rossi et al.

    Gas sensing on unmanned vehicles: challenges and opportunities

    2017 New Generation of CAS (NGCAS)

    (2017)
  • J. Burgués et al.

    Smelling nano aerial vehicle for gas source localization and mapping

    Sensors

    (2019)
  • J. Kennedy

    Zigzagging and casting as a programmed response to wind-borne odour: a review

    Physiol. Entomol.

    (1983)
  • A.J. Rutkowski et al.

    A robotic platform for testing moth-inspired plume tracking strategies

    Proc. IEEE Intern. Conf. on Robotics and Automation (ICRA)

    (2004)
  • W. Li et al.

    Moth-inspired chemical plume tracing on an autonomous underwater vehicle

    IEEE Trans. Robot.

    (2006)
  • X. Kang et al.

    Moth-inspired plume tracing via multiple autonomous vehicles under formation control

    Adapt. Behav.

    (2012)
  • S.I. Azuma et al.

    Stochastic source seeking by mobile robots

    IEEE Trans. Autom. Control

    (2012)
  • S.-J. Liu et al.

    Stochastic Averaging and Stochastic Extremum Seeking

    (2012)
  • S. Li et al.

    Cooperative distributed source seeking by multiple robots: Algorithms and experiments

    IEEE/ASME Trans. Mechatron.

    (2014)
  • J.-B. Masson

    Olfactory searches with limited space perception

    Proc. Natl. Acad. Sci. (PNAS)

    (2013)
  • M. Vergassola et al.

    ‘Infotaxis’ as a strategy for searching without gradients

    Nature

    (2007)
  • Cited by (20)

    • On UAV source seeking with complex dynamic characteristics and multiple constraints: A cooperative standoff monitoring mode

      2022, Aerospace Science and Technology
      Citation Excerpt :

      Compared with ground mobile robots, unmanned aerial vehicles (UAVs) have advantages such as a rapid response, high-cost performance and three-dimensional (3-D) search ability, which make them promising and powerful platforms for pollutant source seeking [3–5]. Generally, existing source-seeking methods can be divided into the three categories of up-flow motion methods, information gain-based methods and gradient-based methods [6]. Additionally, up-flow motion methods, which are referred to as anemotaxis, are inspired by the foraging and mate seeking behaviors of insects such as moths.

    View all citing articles on Scopus
    View full text