You are here

Data mules

volkan-isler's picture

Some sensor network applications (e.g. habitat monitoring) require collecting data from sensors sparsely deployed over a large area. In these scenarios, robots can act as data mules and gather the data from the sensors. We have been working on such systems for a few years now. Here are some highlights:

In [wirelessComm08], we built a proof-of-concept system and showed experimentally that
using robots can yield significant energy savings. More recently, we
studied the following problems that arise in systems where mobile robots periodically collect
data from (static) wireless sensor network nodes. Suppose we are given
approximate locations of the static nodes:

  • In what order should the robots visit the nodes? This sounds like TSP,
    but the problem, which we call the Data Gathering Problem (DGP),
    differs from TSP in the following aspects: In DGP, the objective is to compute
    a tour for each robot in such a way that minimizes the time to collect
    data from all devices. In order to download the data from a device, a
    robot must visit a point within the communication range of the
    device (not necessarily the point itself). Then, it spends a fixed amount of time to download the
    data. Thus, the time to complete a tour depends on not only the travel
    time but also the time to download the data, and the number of devices
    visited along the tour (in TSP, it depends only on the distance).
  • From the static node's perspective: given the stochastic
    nature of the robot's arrival, what is an energy-efficient strategy to
    wake up and send/receive beacon messages? Such a strategy must
    simultaneously minimize the robot's waiting time and the number of
    beacon messages. For this problem, we presented an optimal algorithm.
  • From the robot's perspective: given the
    stochastic nature of the wireless link quality, what is an
    energy-efficient motion strategy to find a good pose (location and
    orientation) from where the data can be downloaded efficiently? The
    robot must be able to find such a location quickly but without taking
    too many measurements so as to conserve the static node's energy.
    When the signal strength function is arbitrary, it is easy to show
    that there is no competitive online strategy for this problem. We present an efficient,
    data-driven heuristic based on experiments.

For the first problem, we will present an approximation algorithm in [iros'09]. The last two results will
appear in our [isrr'09 paper], along with a system implementation for an indoor data collection application.

Currently, we are working (hard!) on building two systems that can be used by field scientists
in environmental monitoring applications. More on this coming soon...