Persistent Monitoring

Introduction

Systems consisting of cooperating mobile agents are often used to perform tasks such as coverage control, surveillance, and environmental sampling. The persistent monitoring problem arises when agents must monitor a dynamically changing environment which cannot be fully covered by a stationary team of agents (as in the coverage control). A result of the exploration process is the eventual discovery of various “points of interest” which, once detected, become “targets” or “data sources” which need to be monitored. This setting arises in multiple application domains ranging from surveillance, environmental monitoring, and energy management down to nano-scale systems tasked to track fluorescent or magnetic particles for the study of dynamic processes in bio-molecular systems and in nano-medical research.

In contrast to a patrolling problem where “every” point in a mission space must be monitored, the problem we address here involves a “finite number” of targets (typically larger than the number of agents) which the agents must cooperatively monitor through periodic visits. We model each target as a queue which value increases if not being covered and decreases if being covered by agents. Our objective is to minimize the accumulated target values over all targets within a given time horizon.

Simulations

In the following video, 1 agent is performing a persistent monitoring task over a 1D mission space.

The second example involves 2 agents and 5 targets over a time horizon of 500 seconds (simulation time).

We further show in [3] that this optimization scheme can be extended to the problem with targets located in a 2D mission space but agents restricted to move on a graph topology consisting of multiple intersecting 1D line segments. These models, for example, streets in an urban setting, corridors in a building, or, more generally, intersecting paths/curves in 2D spaces.

Moreover, for large scale implementations agents can optimize its trajectory based on local information with access only to the state of its neighbors and the IPA gradient can be calculated in a distributed manner except for one event requiring communication from a non-neighbor agent. Ignoring such non-local events will affect the cooperation among agents and results in little loss of accuracy. It can be interpreted as the “price of anarchy” commonly associated with decentralization limiting agent actions to only local information.

Applications

The following video demonstrates an application of Persistent Monitoring in Smart Cities. We control and coordinate the movement of multiple cooperating agents so as to minimize an uncertainty metric associated with a finite number of targets (buildings, stations, intersections).  This application extends our work from a one-dimensional mission space to 1.5D which is a transition to two-dimensional ones.

Thesis & Selected Publications