My research interests span a variety of topics in artificial intelligence and robotics, including (but not limited to): human-robot interaction, spatial language understanding/modeling, socially assistive robotics, computer vision, robot task learning, probabilistic state estimation, motion planning, and multi-robot coordination. This page highlights the research projects that I've worked on both as a Ph.D. student at USC, and also as an M.S./B.S. student at Carnegie Mellon University.

Current and past research projects include:


Spatial Language-Based Human-Robot Interaction Framework

This work presents a novel methodology for autonomous service robots to interpret and follow spatial language instructions, with and without user-specified natural language constraints and/or unvoiced pragmatic constraints. In particular, this work contributes a general computational framework for the representation of dynamic spatial relations (DSRs), including a novel extension to the semantic field model of spatial prepositions, which enables the representation of path prepositions containing both local and global properties.

Our approach allows for robot motion planning and execution of multi-step instruction sequences in real-world continuous domains while providing robustness to sensor noise and environmental uncertainty.

The research and algorithm topics covered by this work include:

  • Real-World Human-Robot Interaction Considerations
    • Unconstrained natural language input
    • Discourse modeling and reference resolution
    • Pragmatic constraints (e.g., safe paths for people and robot, approach behaviors)
    • Object fetch/hand-off behaviors (with PR2 robot)
    • Noisy sensor information/position estimates
  • Interpreting Spatial Object Pick-and-Place Tasks
    • Grasp fields for probabilistic pick-up planning
    • Combined semantic/pragmatic fields for placement planning (e.g., preference for placement on surfaces, away from edges, in drawers, etc.)
    • Integration with Robot Operating System (ROS) and Gazebo 3D simulator
  • Spatial Language-Based HRI Methodology
    • Novel extension of semantic field model to represent dynamic spatial relations involving paths
    • Representation of DSRs of 'to', 'through', and 'around'
    • Probabilistic inference of instruction semantics, and in the associated grounding of noun phrases
    • Modification of A* cost function with semantic fields for path planning under natural language constraints

"Come into the pen. Lift up the object next to Juan. Give him it and then step back outside the pen and wait by the entryway."
"PR2 can you please head to the dinner table and then pick up the water bottle and take it to my desk so I can have a drink later"
Select Videos: [View All Videos]

Select Publications: [View All Publications]

  • Juan Fasola and Maja J. Matarić. "Using Semantic Fields to Model Dynamic Spatial Relations in a Robot Architecture for Natural Language Instruction of Service Robots". In IEEE/RSJ International Conference on Intelligent Robots and Systems 2013 (IROS 2013), Tokyo, Japan, Nov 2013. IROS 2013 CoTeSys Cognitive Robotics Best Paper Award Winner.[PDF]
  • Juan Fasola and Maja J. Matarić. "Using Spatial Semantic and Pragmatic Fields to Interpret Natural Language Pick-and-Place Instructions for a Mobile Service Robot". In International Conference on Social Robotics (ICSR 2013), Bristol, UK, Oct 2013. [PDF]
  • Juan Fasola and Maja J. Matarić. "Modeling Dynamic Spatial Relations with Global Properties for Natural Language-Based Human-Robot Interaction". In 22nd IEEE International Symposium on Robot and Human Interactive Communication (RO-MAN), Gyeongju, Korea, Aug 2013. [PDF]

[Back To Top]


Socially Assistive Robot Exercise Coach for Older Adults

This research work presents the approach, design methodology, and implementation details of a novel socially assistive robot (SAR) system developed to motivate and engage elderly users in simple physical exercise. Our SAR system approach incorporates insights from psychology research into intrinsic motivation and contributes clear design principles developed to maximize the probability of success of SAR-based therapeutic interventions.

To validate our system approach and its effectiveness in gaining user acceptance and motivating physical exercise, three user studies were conducted with older adults, to investigate: 1) the effect of praise and relational discourse in the system towards increasing user motivation; 2) the role of user autonomy and choice within the interaction; and 3) the effect of embodiment in the system by comparing user evaluations of similar physically and virtually embodied SAR exercise coaches in addition to evaluating the overall SAR system.

The research and algorithm topics covered by this work include:

  • Increasing User Intrinsic Motivation
    • SAR-guided therapy design principles
    • Flow state/optimal challenge level
    • Real-time task feedback
    • Positive reinforcement (e.g., praise)
    • Relational behaviors (e.g., politeness, personable, continuation)
    • Competence and trust in human-robot relationship
  • The Effect of Embodiment
    • Physical robot vs. virtual robot (social presence)
  • Real-Time Visual User Activity Recognition (e.g., arm poses)

SAR Exercise Coach interacting with user during one-on-one exercise session
Bandit robot (physical embodiment vs. virtual embodiment)
Real-time visual user activity recognition
algorithm output
Select Videos: [View All Videos]

Select Publications: [View All Publications]

  • Juan Fasola and Maja J. Matarić. "A Socially Assistive Robot Exercise Coach for the Elderly". In Journal of Human-Robot Interaction, 2(2):3-32, Jun 2013. [PDF]
  • Juan Fasola and Maja J. Matarić. "Using Socially Assistive Human-Robot Interaction to Motivate Physical Exercise for Older Adults". In Proceedings of the IEEE - Special Issue on Quality of Life Technology, 100(8):2512-2526, Aug 2012. [PDF]
  • Juan Fasola and Maja J. Matarić. "Robot Exercise Instructor: A Socially Assistive Robot System to Monitor and Encourage Physical Exercise for the Elderly". In 19th IEEE International Symposium in Robot and Human Interactive Communication (Ro-Man 2010), Viareggio, Italy, Sep 2010. Best Paper Award Nominee. [PDF]

[Back To Top]


Real-Time Vision and Artificial Intelligence for Multi-Robot Teams

This work focuses on the design and implementation of real-time artificial intelligence algorithms for multi-robot teams. Specifically, this work contributes computational methods in the RoboCup robot soccer domain for the visual detection of objects, agents, and landmarks, in addition to the development of intelligent robot soccer behaviors with active sensing for increased situational awareness during shooting and passing.

The research and algorithm topics covered by this work include:

  • Real-Time Vision Algorithms
    • Robot detection (teammate vs. opponent)
    • Landmark detection (goals, field lines, markers)
    • Ball detection
  • Artificial Intelligence for Robot Soccer
    • Multi-robot coordination (attacker, supporter, defender)
    • Robot localization (particle filter-based method)
    • World modeling (ball velocity, opponent tracking, etc.)
    • Development of efficient robot walk gate/kick motions

Select Videos: [View All Videos]

Select Publications: [View All Publications]

  • Juan Fasola. "Real-Time Visual Input for Intelligent Robot Soccer Behaviors". Master's Thesis, Carnegie Mellon University, Aug 2007. [PDF]
  • Juan Fasola and Manuela Veloso. "Real-Time Object Detection using Segmented and Grayscale Images". In International Conference on Robotics and Automation (ICRA), Orlando, FL, May 2006. [PDF]
  • Juan Fasola, Paul E. Rybski, and Manuela Veloso. "Fast Goal Navigation with Obstacle Avoidance using a Dynamic Local Visual Model". In VII Brazilian Symposium of Artificial Intelligence (SPAI'05), São Luiz, Brazil, Sep 2005. [PDF]

[Back To Top]

© 2014 Juan Fasola. All rights reserved.