Active Information-Based Localization and Mapping
Philip Dames, Benjamin Charrow, Sikang Liu, Nathan Michael, Vijay Kumar

Citation
Philip Dames, Benjamin Charrow, Sikang Liu, Nathan Michael, Vijay Kumar. "Active Information-Based Localization and Mapping". Talk or presentation, 30, October, 2014; Poster presented at the 2014 TerraSwarm Annual Meeting.

Abstract
In this work we seek to enable a team of mobile robots to autonomously gather information in an environment to relay to a remote human operator. We consider two key problems: 1) detecting and localizing an uknown number of objects of interest within a known environment and 2) generating a 3D occupancy grid of an unknown environment using a heterogeneous team of ground and aerial robots. We develop computationally efficient control policies for active perception that incorporate explicit models of sensing and mobility. Like previous work, our policy maximizes an information-theoretic objective function between the distribution over the targets/map and future measurements that can be made by mobile sensors. While most previous methods adopt a myopic, gradient-following approach that yields poor convergence properties, our algorithm searches over a set of paths generated over a range of length scales and is less susceptible to local minima. In doing so, we explicitly incorporate models of sensors, and model the dependence (and independence) of measurements over multiple time steps in a path. Because we consider models of sensing and mobility, our method naturally applies to both ground and aerial vehicles. We present results from simulated and hardware experiments that demonstrate the effectiveness of the proposed control policy.

Electronic downloads

Internal. This publication has been marked by the author for TerraSwarm-only distribution, so electronic downloads are not available without logging in.

Citation formats

• HTML

  Philip Dames, Benjamin Charrow, Sikang Liu, Nathan Michael,
  Vijay Kumar. <a
  href="http://www.terraswarm.org/pubs/451.html"><i>Active
  Information-Based Localization and
  Mapping</i></a>, Talk or presentation,  30,
  October, 2014; Poster presented at the <a
  href="http://www.terraswarm.org/conferences/14/annual"
  >2014 TerraSwarm Annual Meeting</a>.

• Plain text

  Philip Dames, Benjamin Charrow, Sikang Liu, Nathan Michael,
  Vijay Kumar. "Active Information-Based Localization and
  Mapping". Talk or presentation,  30, October, 2014;
  Poster presented at the <a
  href="http://www.terraswarm.org/conferences/14/annual"
  >2014 TerraSwarm Annual Meeting</a>.

• BibTeX

  @presentation{DamesCharrowLiuMichaelKumar14_ActiveInformationBasedLocalizationMapping,
      author = {Philip Dames and Benjamin Charrow and Sikang Liu
                and Nathan Michael and Vijay Kumar},
      title = {Active Information-Based Localization and Mapping},
      day = {30},
      month = {October},
      year = {2014},
      note = {Poster presented at the <a
                href="http://www.terraswarm.org/conferences/14/annual"
                >2014 TerraSwarm Annual Meeting</a>.},
      abstract = {In this work we seek to enable a team of mobile
                robots to autonomously gather information in an
                environment to relay to a remote human operator.
                We consider two key problems: 1) detecting and
                localizing an uknown number of objects of interest
                within a known environment and 2) generating a 3D
                occupancy grid of an unknown environment using a
                heterogeneous team of ground and aerial robots. We
                develop computationally efficient control policies
                for active perception that incorporate explicit
                models of sensing and mobility. Like previous
                work, our policy maximizes an
                information-theoretic objective function between
                the distribution over the targets/map and future
                measurements that can be made by mobile sensors.
                While most previous methods adopt a myopic,
                gradient-following approach that yields poor
                convergence properties, our algorithm searches
                over a set of paths generated over a range of
                length scales and is less susceptible to local
                minima. In doing so, we explicitly incorporate
                models of sensors, and model the dependence (and
                independence) of measurements over multiple time
                steps in a path. Because we consider models of
                sensing and mobility, our method naturally applies
                to both ground and aerial vehicles. We present
                results from simulated and hardware experiments
                that demonstrate the effectiveness of the proposed
                control policy.},
      URL = {http://terraswarm.org/pubs/451.html}
  }
  

Posted by Philip Dames on 7 Nov 2014.
Groups: services