Citation
Ilge Akkaya. "Building compositional learning and optimization applications for mobile sensor networks with PILOT". Talk or presentation, 16, October, 2015; Presented at the Eleventh Biennial Ptolemy Miniconference, Berkeley.

Abstract
Emerging distributed cyber-physical systems (CPS) are integrating a variety of sensors, actuators, communication networks, and computation resources, giving rise to a highly complex and heterogeneous system architecture. Building integrated applications that involve estimation, inference, machine learning, and coordination is becoming a difficult task due to the heterogeneous na.ture of involved components. We present PILOT (Ptolemy Inference, Learning, and Optimization Toolkit), an actor-oriented machine learning and optimization toolkit that is designed for developing data intensive distributed applications for heterogeneous distributed CPS. PILOT gives clear semantics to the learning and optimization workflow and leverages compositionality in this framework. Moreover, the actor interfaces and algorithmic abstractions provide accessibility to system designers who are non-experts in machine learning. The toolkit builds upon the aspect-oriented modeling (AOM) paradigm to provide compositional interfaces and separation.of-concerns during the development process and enables sensed environmental constraints to be dynamically integrated into learning and optimization tasks in run time via the accessor framework. We showcase PILOT on a simulation based robotic disaster response scenario, where a heterogeneous team of cooperative robots carry out a simultaneous localization and mapping task in an unknown environment. PILOT enables mobile target localization with map constraints in such scenario, where the map data is modified in real time via an accessor interface. Using a variety of heterogeneous sensor inputs (e.g., microphones, rangefinders, cameras), the learning application can perform target state estimation and path planning given the most up-to-date mapping information obtained from the robots.

Electronic downloads

• Akkaya_PILOT_PtolemyMiniConference_16Oct2015.pdf · application/pdf · 12979 kbytes

• HTML

  Ilge Akkaya. <a
  href="http://chess.eecs.berkeley.edu/pubs/1124.html"><i>Building
  compositional learning and optimization applications for
  mobile sensor networks with PILOT</i></a>, Talk
  or presentation,  16, October, 2015; Presented at the <a
  href="http://ptolemy.eecs.berkeley.edu/conferences/15/"
  >Eleventh Biennial Ptolemy Miniconference</a>,
  Berkeley.

• Plain text

  Ilge Akkaya. "Building compositional learning and
  optimization applications for mobile sensor networks with
  PILOT". Talk or presentation,  16, October, 2015;
  Presented at the <a
  href="http://ptolemy.eecs.berkeley.edu/conferences/15/"
  >Eleventh Biennial Ptolemy Miniconference</a>,
  Berkeley.

• BibTeX

  @presentation{Akkaya15_BuildingCompositionalLearningOptimizationApplications,
      author = {Ilge Akkaya},
      title = {Building compositional learning and optimization
                applications for mobile sensor networks with PILOT},
      day = {16},
      month = {October},
      year = {2015},
      note = {Presented at the <a
                href="http://ptolemy.eecs.berkeley.edu/conferences/15/"
                >Eleventh Biennial Ptolemy Miniconference</a>,
                Berkeley},
      abstract = {Emerging distributed cyber-physical systems (CPS)
                are integrating a variety of sensors, actuators,
                communication networks, and computation resources,
                giving rise to a highly complex and heterogeneous
                system architecture. Building integrated
                applications that involve estimation, inference,
                machine learning, and coordination is becoming a
                difficult task due to the heterogeneous na.ture of
                involved components. We present PILOT (Ptolemy
                Inference, Learning, and Optimization Toolkit), an
                actor-oriented machine learning and optimization
                toolkit that is designed for developing data
                intensive distributed applications for
                heterogeneous distributed CPS. PILOT gives clear
                semantics to the learning and optimization
                workflow and leverages compositionality in this
                framework. Moreover, the actor interfaces and
                algorithmic abstractions provide accessibility to
                system designers who are non-experts in machine
                learning. The toolkit builds upon the
                aspect-oriented modeling (AOM) paradigm to provide
                compositional interfaces and
                separation.of-concerns during the development
                process and enables sensed environmental
                constraints to be dynamically integrated into
                learning and optimization tasks in run time via
                the accessor framework. We showcase PILOT on a
                simulation based robotic disaster response
                scenario, where a heterogeneous team of
                cooperative robots carry out a simultaneous
                localization and mapping task in an unknown
                environment. PILOT enables mobile target
                localization with map constraints in such
                scenario, where the map data is modified in real
                time via an accessor interface. Using a variety of
                heterogeneous sensor inputs (e.g., microphones,
                rangefinders, cameras), the learning application
                can perform target state estimation and path
                planning given the most up-to-date mapping
                information obtained from the robots. },
      URL = {http://chess.eecs.berkeley.edu/pubs/1124.html}
  }
  

Posted by Christopher Brooks on 19 Oct 2015.
Groups: ptolemy
For additional information, see the Publications FAQ or contact webmaster at chess eecs berkeley edu.