Intention Recognition

From WWWwikiEn

SFB 588 - Kitchen Setting.

Human-robot cooperation is governed by mutual estimation of intentions. Providing intention recognition capabilities to technical systems will enable these to communicate implicitly with their human users. This research is funded by the Collaborative Research Center 588 "Humanoid Robots - Learning and Cooperating Multimodal Robots", established on the 1st of July 2001 by the Deutsche Forschungsgemeinschaft. ISAS is a member of this center since 2004.

Intention recognition at the ISAS is based on a stochastic modeling approach using hybrid dynamic bayesian networks. Bayesian networks are cascaded stochastic models and represent causal relations between situations, intentions, actions and observations explicitily. The focus of our research is on the representation, inference, and learning of non-linear dependencies and hybrid scenarios, which contain continuous as well as discrete random variables.

As a baseline basic human intention recognition skills were quantified in experimental studies in a kitchen setting. On the basis of these experiments an estimator for intention recognition was developed. This estimator achieves recognition results comparable to those of human users in the given setting.

Scientist(s): Simon Friedberger, Peter Krauthausen.
Alumni & Collaborator(s): Oliver Schrempf, Marcus Baum.

Selected Publications

At ISAS we propose the use of a generic model of the human behavior for intention recognition
of varying scales/scopes and complexity.

Oliver C. Schrempf, Uwe D. Hanebeck,
A Generic Model for Estimating User Intentions in Human-Robot Cooperation,

Proceedings of the 2nd International Conference on Informatics in Control, Automation and Robotics (ICINCO 2005), 3:251-256, Barcelona, Spain, September, 2005.

Author : Oliver C. Schrempf, Uwe D. Hanebeck
Title : A Generic Model for Estimating User Intentions in Human-Robot Cooperation
In : Proceedings of the 2nd International Conference on Informatics in Control, Automation and Robotics (ICINCO 2005)
Date : September 2005

Abstract

The recognition of user intentions is an important feature for humanoid
robots to make implicit and human-like interactions possible. In
this paper, we introduce a formal view on user-intentions in human-machine
interaction and how they can be estimated by observing user actions.
We use Hybrid Dynamic Bayesian Networks to develop a generic model
that includes connections between intentions, actions, and sensor
measurements. This model can be used to extend arbitrary human-machine
applications by intention recognition.

We developed Hybrid Dynamic Bayesian Networks that allow for nonlinear dependencies
between the variables and closed-form inference. This is very convenient for modeling mixed discrete and
continuous valued domains which are common in the human surrounding.

Oliver C. Schrempf, Uwe D. Hanebeck,
A New Approach for Hybrid Bayesian Networks Using Full Densities,

Proceedings of the 6th International Workshop on Computer Science and Information Technologies (CSIT 2004), 1:32-37, Budapest, Hungary, October, 2004.

Author : Oliver C. Schrempf, Uwe D. Hanebeck
Title : A New Approach for Hybrid Bayesian Networks Using Full Densities
In : Proceedings of the 6th International Workshop on Computer Science and Information Technologies (CSIT 2004)
Date : October 2004

Abstract

In this article, a new mechanism is described for modeling and evaluating
hybrid Bayesian networks. The approach uses Gaussian mixtures and
Dirac mixtures as messages to calculate marginal densities. The mechanism
is proven to be exact, hence the accuracy of resulting marginals
is only dependending on the accuracy of the conditional densities.
As these densities are approximated by means of Gaussian mixtures,
any desired precision can be achieved. The presented approach removes
the restrictions concerning the ancestry of discrete nodes often
made in literature. Hence it enables the designer to model arbitrary
parent-child relationships using continuous and discrete variables.

Oliver C. Schrempf, Uwe D. Hanebeck,
Evaluation of Hybrid Bayesian Networks using Analytical Density Representations,

Proceedings of the 16th IFAC World Congress (IFAC 2005), Prague, Czech Republic, July, 2005.

Author : Oliver C. Schrempf, Uwe D. Hanebeck
Title : Evaluation of Hybrid Bayesian Networks using Analytical Density Representations
In : Proceedings of the 16th IFAC World Congress (IFAC 2005)
Date : July 2005

Abstract

In this article, a new mechanism is described for modeling and evaluating
Hybrid Dynamic Bayesian networks. The approach uses Gaussian mixtures
and Dirac mixtures as messages to calculate marginal densities. As
these densities are approximated by means of Gaussian mixtures, any
desired precision is possible. The presented approach removes the
restrictions of sample based evaluation of Bayesian networks since
it uses an analytical approximation scheme for probability densities
which systematically minimizes the distance between the exact and
the approximate density.

Additionally, methods for learning the parameters of Hybrid Bayesian Networks and
learning continuous nonlinear dependencies based on sparse kernel (conditional) densities were developed.

Anne Hanselmann, Oliver C. Schrempf, Uwe D. Hanebeck,
Optimal Parametric Density Estimation by Minimizing an Analytic Distance Measure,

Proceedings of the 10th International Conference on Information Fusion (Fusion 2007), Quebec, Canada, July, 2007.

Author : Anne Hanselmann, Oliver C. Schrempf, Uwe D. Hanebeck
Title : Optimal Parametric Density Estimation by Minimizing an Analytic Distance Measure
In : Proceedings of the 10th International Conference on Information Fusion (Fusion 2007)
Date : July 2007

Abstract

In this paper, we present a novel approach to parametric density estimation
from given samples. The samples are treated as a parametric density
function by means of a Dirac mixture, which allows for applying analytic
optimization techniques. The method is based on minimizing a distance
measure between the integral of the approximation function and the
empirical cumulative distribution function (EDF) of the given samples,
where the EDF is represented by the integral of the Dirac mixture.
Since this minimization problem cannot be solved directly in general,
a progression technique is applied. Increased performance of the
approach in comparison to iterative maximum likelihood approaches
is shown in simulations.

In order to scale the proposed modeling approach to large environments,
we propose the use of situation-specific inference.

Other developments include a method for calculating hybrid clusters, a complexity reduction
for restricted types of systems or methods for data association and knowledge modeling.

Oliver C. Schrempf, David Albrecht, Uwe D. Hanebeck,
Tractable Probabilistic Models for Intention Recognition Based on Expert Knowledge,

Proceedings of the 2007 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2007), pp. 1429-1434, San Diego, California, USA, November, 2007.

Author : Oliver C. Schrempf, David Albrecht, Uwe D. Hanebeck
Title : Tractable Probabilistic Models for Intention Recognition Based on Expert Knowledge
In : Proceedings of the 2007 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2007)
Date : November 2007

Abstract

Intention recognition is an important topic in human-robot cooperation
that can be tackled using probabilistic model-based methods. A popular
instance of such methods are Bayesian networks where the dependencies
between random variables are modeled by means of a directed graph.
Bayesian networks are very efficient for treating networks with conditionally
independent parts. Unfortunately, such independence sometimes has
to be constructed by introducing so called hidden variables with
an intractably large state space. An example are human actions which
depend on human intentions and on other human actions. Our goal in
this paper is to find models for intention-action mapping with a
reduced state space in order to allow for tractable on-line evaluation.
We present a systematic derivation of the reduced model and experimental
results of recognizing the intention of a real human in a virtual
environment.

Oliver C. Schrempf, Anne Hanselmann, Uwe D. Hanebeck,
Efficient Representation and Fusion of Hybrid Joint Densities for Clusters in Nonlinear Hybrid Bayesian Networks,

Proceedings of the 9th International Conference on Information Fusion (Fusion 2006), Florence, Italy, July, 2006.

Author : Oliver C. Schrempf, Anne Hanselmann, Uwe D. Hanebeck
Title : Efficient Representation and Fusion of Hybrid Joint Densities for Clusters in Nonlinear Hybrid Bayesian Networks
In : Proceedings of the 9th International Conference on Information Fusion (Fusion 2006)
Date : July 2006

Abstract

Undirected cycles in Bayesian networks are often treated by using
clustering methods. This results in networks with nodes characterized
by joint probability densities instead of marginal densities. An
efficient representation of these hybrid joint densities is essential
especially in nonlinear hybrid networks containing continuous as
well as discrete variables. In this article we present a unified
representation of continuous, discrete, and hybrid joint densities.
This representation is based on Gaussian and Dirac mixtures and allows
for analytic evaluation of arbitrary hybrid networks without loosing
structural information, even for networks containing clusters. Furthermore
we derive update formulae for marginal and joint densities from a
system theoretic point of view by treating a Bayesian network as
a system of cascaded subsystems. Together with the presented mixture
representation of densities this yields an exact analytic updating
scheme.