Citation
Ivan Papusha, Eugene Lavretsky, Richard Murray. "Collaborative System Identification via Parameter Consensus". Talk or presentation, 5, November, 2013; Poster presented at the 2013 TerraSwarm Annual Meeting.

Abstract
Standard schemes in system identification and adaptive control rely on persistence of excitation to guarantee parameter convergence. Inspired by networked systems, we extend parameter adaptation to the multi-agent setting by combining a gradient law with consensus dynamics. The gradient law introduces a learning signal, while consensus dynamics preferentially push each agent's parameter estimates toward those of its neighbors. We show that the resulting online, decentralized parameter estimator combines local and neighboring information to identify the true parameters even if no single agent employs a persistently exciting input. We also elaborate upon collective persistence of excitation in networked adaptive algorithms.

Electronic downloads

• http://www.cds.caltech.edu/~murray/preprints/plm14-acc.pdf

• HTML

  Ivan Papusha, Eugene Lavretsky, Richard Murray. <a
  href="http://www.terraswarm.org/pubs/172.html"><i>Collaborative
  System Identification via Parameter
  Consensus</i></a>, Talk or presentation,  5,
  November, 2013; Poster presented at the <a
  href="http://www.terraswarm.org/conferences/13/annual"
  >2013 TerraSwarm Annual Meeting</a>.

• Plain text

  Ivan Papusha, Eugene Lavretsky, Richard Murray.
  "Collaborative System Identification via Parameter
  Consensus". Talk or presentation,  5, November, 2013;
  Poster presented at the <a
  href="http://www.terraswarm.org/conferences/13/annual"
  >2013 TerraSwarm Annual Meeting</a>.

• BibTeX

  @presentation{PapushaLavretskyMurray13_CollaborativeSystemIdentificationViaParameterConsensus,
      author = {Ivan Papusha and Eugene Lavretsky and Richard
                Murray},
      title = {Collaborative System Identification via Parameter
                Consensus},
      day = {5},
      month = {November},
      year = {2013},
      note = {Poster presented at the <a
                href="http://www.terraswarm.org/conferences/13/annual"
                >2013 TerraSwarm Annual Meeting</a>.},
      abstract = {Standard schemes in system identification and
                adaptive control rely on persistence of excitation
                to guarantee parameter convergence. Inspired by
                networked systems, we extend parameter adaptation
                to the multi-agent setting by combining a gradient
                law with consensus dynamics. The gradient law
                introduces a learning signal, while consensus
                dynamics preferentially push each agent's
                parameter estimates toward those of its neighbors.
                We show that the resulting online, decentralized
                parameter estimator combines local and neighboring
                information to identify the true parameters even
                if no single agent employs a persistently exciting
                input. We also elaborate upon collective
                persistence of excitation in networked adaptive
                algorithms.},
      URL = {http://terraswarm.org/pubs/172.html}
  }
  

Posted by Ivan Papusha on 4 Nov 2013.