Robotics Research Lab
USC Computer Science
USC Engineering
/ Research / Projects / People-Aware Navigation For Goal-Oriented Behavior Involving a Human Partner

Motivation Approach
Publications Support Contact Details

In order to facilitate effective autonomous interaction behavior for human-robot interaction the robot should be able to execute goal-oriented behavior while reacting to sensor feedback related to the people with which it is interacting. Prior work has demonstrated that autonomously sensed distance-based features can be used to correctly detect user state. We wish to demonstrate that such models can also be used to weight action selection as well. This paper considers the problem of moving to a goal along with a partner, demonstrating that a learned model can be used to weight trajectories of a navigation system for autonomous movement. This project presents a realization of a person-aware navigation system which requires no ad-hoc parameter tuning, and no input other than a small set of training examples. This system is currently validated using an in-lab demonstration of people-aware navigation using the described system.


Our motivating problem is an example of inappropriate social behavior. We observed during our preliminary work using robots with children with Autism Spectrum Disorders (ASD) that when the child was socially interacting with the robot, when the robot moved away from the child and toward some other goal, the child would not follow. It appeared from an observer's perspective that the robot was ignoring the child. If the robot's navigation better reflected appropriate social behavior, a child would be more likely to follow the robot. Prior work has demonstrated that spatial movement behavior can be described using a Gaussian Mixture Model (GMM) and used to detect the social behavior of a person interacting with the robot. If such a model can be used to detect social behavior occurring between person and robot, then such a model might also be used in action selection.

The focus of this work is to develop a system which can exhibit goal-oriented behavior while using people-related sensing as part of the planning process. We have chosen to explore the task of moving to a goal while being accompanied by a partner. An explicit property of such an action is that both the robot and partner arrive at the goal, while proximity between robot and partner is also a consideration. What is not explicit is the relationship between goal distance and partner distance. What is the appropriate social distance between partner and robot? Does that change with the progress toward the goal? Given a set of example movements, where a partner is following a robot toward a goal, we wish to model proper follow behavior. Can such a data-driven model be used for trajectory planning?

We use the Willow Garage PR2 robot in concert with our overhead camera system.


A larger list of relevant publications can be found here.

This project is funded in part by the Okawa Foundation, the Institute for Creative Technologies, the Dan Marino Foundation through the Marino Autism Research Institute (MARI), the USC Provost's Center for Interdisciplinary Research, and AnthroTronix, Inc.


David Feil-Seifer