Multimodal Dialogue Research Articles

A Multimodal Dialog System for Language Assessment: Current State and Future Directions

We present work in progress on a multimodal dialog system for English language assessment using a modular cloud-based architecture adhering to open industry standards. Among the modules being developed for the system, multiple modules heavily exploit machine learning techniques, including speech recognition, spoken language proficiency rating, speaker recognition, and the scoring of behaviors in multimodal data streams.

Using Vision and Speech Features for Automated Prediction of Performance Metrics in Multimodal Dialogs

Predicting and analyzing multimodal dialog user experience (UX) metrics, such as overall call experience, caller engagement, and latency, among other metrics, in an ongoing manner is important for evaluating such systems. We investigate automated prediction of multiple such metrics collected from crowdsourced interactions with an open-source, cloud-based multimodal dialog system in the educational domain. We extract features from both the audio and video signals and examine the efficacy of multiple machine learning algorithms in predicting these performance metrics. The best performing audio features consist of multiple low-level audio descriptors—intensity, loudness, cepstra, pitch, and so on—and their functionals, extracted using the OpenSMILE toolkit, while the video features are bags of visual words that use 3D Scale-Invariant Feature Transform descriptors. We find that our proposed methods outperform the majority vote classification baseline in predicting various UX metrics rated by both the user and experts. Our results suggest that such automated prediction of performance metrics can not only inform the qualitative and quantitative analysis of dialogs but also be potentially incorporated into dialog management routines for positively impacting UX and other metrics during the course of the interaction.

Constructionist Design Methodology for Interactive Intelligences

We present a methodology for designing and implementing interactive intelligences. The constructionist design methodology (CDM) -- so called because it advocates modular building blocks and incorporation of prior work -- addresses factors that we see as key to future advances in AI, including support for interdisciplinary collaboration, coordination of teams, and large-scale systems integration. We test the methodology by building an interactive multifunctional system with a real-time perception- action loop. The system, whose construction relied entirely on the methodology, consists of an embodied virtual agent that can perceive both real and virtual objects in an augmented-reality room and interact with a user through coordinated gestures and speech. Wireless tracking technologies give the agent awareness of the environment and the user's speech and communicative acts. User and agent can communicate about things in the environment, their placement, and their function, as well as about more abstract topics, such as current news, through situated multimodal dialogue. The results demonstrate the CDM's strength in simplifying the modeling of complex, multifunctional systems that require architectural experimentation and exploration of unclear subsystem boundaries, undefined variables, and tangled data flow and control hierarchies.

Multimodal control system for autonomous vehicles using speech and gesture recognition

The recent development of autonomous vehicles has attracted much attention, but operating these vehicles may be too complex for average users. Therefore, we propose an intuitive, multimodal interface for the control of autonomous vehicles using speech and gesture recognition to interpret and execute the commands of users. For example, if the user says “turn there” while pointing at a landmark, the vehicle can utilize this behavior to correctly understand and comply with the user’s intent. To achieve this, we designed a two-part interface consisting of a multimodal understanding component and a dialog control component. Our multimodal understanding and dialog control components can be seen as a concatenation of two separate transducers. One transducer is used for multimodal understanding and the other for a conventional dialog system. We then construct a combined transducer from these two transducers. We developed various scenarios which might arise while operating an autonomous vehicle and displayed these...

Artificial cognition for social human–robot interaction: An implementation

Human–Robot Interaction challenges Artificial Intelligence in many regards: dynamic, partially unknown environments that were not originally designed for robots; a broad variety of situations with rich semantics to understand and interpret; physical interactions with humans that requires fine, low-latency yet socially acceptable control strategies; natural and multi-modal communication which mandates common-sense knowledge and the representation of possibly divergent mental models. This article is an attempt to characterise these challenges and to exhibit a set of key decisional issues that need to be addressed for a cognitive robot to successfully share space and tasks with a human.We identify first the needed individual and collaborative cognitive skills: geometric reasoning and situation assessment based on perspective-taking and affordance analysis; acquisition and representation of knowledge models for multiple agents (humans and robots, with their specificities); situated, natural and multi-modal dialogue; human-aware task planning; human–robot joint task achievement. The article discusses each of these abilities, presents working implementations, and shows how they combine in a coherent and original deliberative architecture for human–robot interaction. Supported by experimental results, we eventually show how explicit knowledge management, both symbolic and geometric, proves to be instrumental to richer and more natural human–robot interactions by pushing for pervasive, human-level semantics within the robot's deliberative system.

Multimodal Dialogues: Making Visual Meaning through the Translation of Tacit Knowledge

Multimodal Dialogues: Making Visual Meaning through the Translation of Tacit Knowledge

The roles and recognition of Haptic-Ostensive actions in collaborative multimodal human–human dialogues

The RoboHelper project has the goal of developing assistive robots for the elderly. One crucial component of such a robot is a multimodal dialogue architecture, since collaborative task-oriented human–human dialogue is inherently multimodal. In this paper, we focus on a specific type of interaction, Haptic-Ostensive (H-O) actions, that are pervasive in collaborative dialogue. H-O actions manipulate objects, but they also often perform a referring function.We collected 20 collaborative task-oriented human–human dialogues between a helper and an elderly person in a realistic setting. To collect the haptic signals, we developed an unobtrusive sensory glove with pressure sensors. Multiple annotations were then conducted to build the Find corpus. Supervised machine learning was applied to these annotations in order to develop reference resolution and dialogue act classification modules. Both corpus analysis, and these two modules show that H-O actions play a crucial role in interaction: models that include H-O actions, and other extra-linguistic information such as pointing gestures, perform better.For true human–robot interaction, all communicative intentions must of course be recognized in real time, not on the basis of annotated categories. To demonstrate that our corpus analysis is not an end in itself, but can inform actual human–robot interaction, the last part of our paper presents additional experiments on recognizing H-O actions from the haptic signals measured through the sensory glove. We show that even though pressure sensors are relatively imprecise and the data provided by the glove is noisy, the classification algorithms can successfully identify actions of interest within subjects.

A Natural Conversational Virtual Human with Multimodal Dialog System

The making of virtual human character to be realistic and credible in real time automated dialog animation system is necessary. This kind of animation carries importance elements for many applications such as games, virtual agents and movie animations. It is also considered important for applications which require interaction between human and computer. However, for this purpose, it is compulsory that the machine should have sufficient intelligence for recognizing and synthesizing human voices. As one of the most vital interaction method between human and machine, speech has recently received significant attention, especially in avatar research innovation. One of the challenges is to create precise lip movements of the avatar and synchronize it with a recorded audio. This paper specifically introduces the innovative concept of multimodal dialog systems of the virtual character and focuses the output part of such systems. More specifically, its focus is on behavior planning and developing the data control languages (DCL).

Fluent Human–Robot Dialogues About Grounded Objects in Home Environments

To provide a spoken interaction between robots and human users, an internal representation of the robots sensory information must be available at a semantic level and accessible to a dialogue system in order to be used in a human-like and intuitive manner. In this paper, we integrate the fields of perceptual anchoring (which creates and maintains the symbol-percept correspondence of objects) in robotics with multimodal dialogues in order to achieve a fluent interaction between humans and robots when talking about objects. These everyday objects are located in a so-called symbiotic system where humans, robots, and sensors are co-operating in a home environment. To orchestrate the dialogue system, the IrisTK dialogue platform is used. The IrisTK system is based on modelling the interaction of events, between different modules, e.g. speech recognizer, face tracker, etc. This system is running on a mobile robot device, which is part of a distributed sensor network. A perceptual anchoring framework, recognizes objects placed in the home and maintains a consistent identity of the objects consisting of their symbolic and perceptual data. Particular effort is placed on creating flexible dialogues where requests to objects can be made in a variety of ways. Experimental validation consists of evaluating the system when many objects are possible candidates for satisfying these requests.

SIMONA – the Slovak embodied conversational agent

The proposed paper deals with the design and development of SIMONA – the Slovak embodied conversational agent. It was created as an extension of a multimodal dialogue system with the BML realizer Elckerlyc as an embodied agent. Elckerlyc is a BML com

MULTIMODAL FUSION AS COMMUNICATIVE ACTS DURING HUMAN–ROBOT INTERACTION

Research on dialog systems is a very active area in social robotics. During the last two decades, these systems have evolved from those based only on speech recognition and synthesis to the current and modern systems, which include new components and multimodality. By multimodal dialogue we mean the interchange of information among several interlocutors, not just using their voice as the mean of transmission but also all the available channels such as gestures, facial expressions, touch, sounds, etc. These channels add information to the message to be transmitted in every dialogue turn. The dialogue manager (IDiM) is one of the components of the robotic dialog system (RDS) and is in charge of managing the dialogue flow during the conversational turns. In order to do that, it is necessary to coherently treat the inputs and outputs of information that flow by different communication channels: audio, vision, radio frequency, touch, etc. In our approach, this multichannel input of information is temporarily fused into communicative acts (CAs). Each CA groups the information that flows through the different input channels into the same pack, transmitting a unique message or global idea. Therefore, this temporary fusion of information allows the IDiM to abstract from the channels used during the interaction, focusing only on the message, not on the way it is transmitted. This article presents the whole RDS and the description of how the multimodal fusion of information is made as CAs. Finally, several scenarios where the multimodal dialogue is used are presented.

Gaze awareness in conversational agents

In face-to-face conversations, speakers are continuously checking whether the listener is engaged in the conversation, and they change their conversational strategy if the listener is not fully engaged. With the goal of building a conversational agent that can adaptively control conversations, in this study we analyze listener gaze behaviors and develop a method for estimating whether a listener is engaged in the conversation on the basis of these behaviors. First, we conduct a Wizard-of-Oz study to collect information on a user's gaze behaviors. We then investigate how conversational disengagement, as annotated by human judges, correlates with gaze transition, mutual gaze (eye contact) occurrence, gaze duration, and eye movement distance. On the basis of the results of these analyses, we identify useful information for estimating a user's disengagement and establish an engagement estimation method using a decision tree technique. The results of these analyses show that a model using the features of gaze transition, mutual gaze occurrence, gaze duration, and eye movement distance provides the best performance and can estimate the user's conversational engagement accurately. The estimation model is then implemented as a real-time disengagement judgment mechanism and incorporated into a multimodal dialog manager in an animated conversational agent. This agent is designed to estimate the user's conversational engagement and generate probing questions when the user is distracted from the conversation. Finally, we evaluate the engagement-sensitive agent and find that asking probing questions at the proper times has the expected effects on the user's verbal/nonverbal behaviors during communication with the agent. We also find that our agent system improves the user's impression of the agent in terms of its engagement awareness, behavior appropriateness, conversation smoothness, favorability, and intelligence.

A graphical editor for the SMUIML multimodal user interaction description language

We present the results of an investigation on software support for the SMUIML multimodal user interaction description language (UIDL). In particular, we introduce a graphical UIDL editor for the creation of SMUIML scripts. The data management as well as the dialogue modelling in our graphical editor is based on the SMUIML language. Due to the event-centred nature of SMUIML, the multimodal dialogue modelling is represented by a state machine in the graphical SMUIML dialogue editor. Our editor further offers a real-time graphical debugging tool. Compared to existing multimodal dialogue editing solutions, the SMUIML graphical editor offers synchronised dual editing in graphical and textual forms as well as a number of operators for the temporal combination of modalities. The presented graphical editor represents the third component of a triad of tools for the development of multimodal user interfaces, consisting of an XML-based modelling language, a framework for the authoring of multimodal interfaces and a graphical editor.

Multi-modal Dialogue Analysis on Turn-taking and Difference of Variations in Brainwave between Cooperative Learners

Multi-modal Dialogue Analysis on Turn-taking and Difference of Variations in Brainwave between Cooperative Learners

User Localization During Human-Robot Interaction

This paper presents a user localization system based on the fusion of visual information and sound source localization, implemented on a social robot called Maggie. One of the main requisites to obtain a natural interaction between human-human and human-robot is an adequate spatial situation between the interlocutors, that is, to be orientated and situated at the right distance during the conversation in order to have a satisfactory communicative process. Our social robot uses a complete multimodal dialog system which manages the user-robot interaction during the communicative process. One of its main components is the presented user localization system. To determine the most suitable allocation of the robot in relation to the user, a proxemic study of the human-robot interaction is required, which is described in this paper. The study has been made with two groups of users: children, aged between 8 and 17, and adults. Finally, at the end of the paper, experimental results with the proposed multimodal dialog system are presented.

Building Autonomous Sensitive Artificial Listeners

This paper describes a substantial effort to build a real-time interactive multimodal dialogue system with a focus on emotional and nonverbal interaction capabilities. The work is motivated by the aim to provide technology with competences in perceiving and producing the emotional and nonverbal behaviors required to sustain a conversational dialogue. We present the Sensitive Artificial Listener (SAL) scenario as a setting which seems particularly suited for the study of emotional and nonverbal behavior since it requires only very limited verbal understanding on the part of the machine. This scenario allows us to concentrate on nonverbal capabilities without having to address at the same time the challenges of spoken language understanding, task modeling, etc. We first report on three prototype versions of the SAL scenario in which the behavior of the Sensitive Artificial Listener characters was determined by a human operator. These prototypes served the purpose of verifying the effectiveness of the SAL scenario and allowed us to collect data required for building system components for analyzing and synthesizing the respective behaviors. We then describe the fully autonomous integrated real-time system we created, which combines incremental analysis of user behavior, dialogue management, and synthesis of speaker and listener behavior of a SAL character displayed as a virtual agent. We discuss principles that should underlie the evaluation of SAL-type systems. Since the system is designed for modularity and reuse and since it is publicly available, the SAL system has potential as a joint research tool in the affective computing research community.

End-user programming of a social robot by dialog

One of the main challenges faced by social robots is how to provide intuitive, natural and enjoyable usability for the end-user. In our ordinary environment, social robots could be important tools for education and entertainment (edutainment) in a variety of ways. This paper presents a Natural Programming System (NPS) that is geared to non-expert users. The main goal of such a system is to provide an enjoyable interactive platform for the users to build different programs within their social robot platform. The end-user can build a complex net of actions and conditions (a sequence) in a social robot via mixed-initiative dialogs and multimodal interaction. The system has been implemented and tested in Maggie, a real social robot with multiple skills, conceived as a general HRI researching platform. The robot’s internal features (skills) have been implemented to be verbally accessible to the end-user, who can combine them into others that are more complex following a bottom-up model. The built sequence is internally implemented as a Sequence Function Chart (SFC), which allows parallel execution, modularity and re-use. A multimodal Dialog Manager System (DMS) takes charge of keeping the coherence of the interaction. This work is thought for bringing social robots closer to non-expert users, who can play the game of “teaching how to do things” with the robot.

Using speech to identify gesture pen strokes in collaborative, multimodal device descriptions

AbstractOne challenge in building collaborative design tools that use speech and sketch input is distinguishing gesture pen strokes from those representing device structure, that is, object strokes. In previous work, we developed a gesture/object classifier that uses features computed from the pen strokes and the speech aligned with them. Experiments indicated that the speech features were the most important for distinguishing gestures, thus indicating the critical importance of the speech–sketch alignment. Consequently, we have developed a new alignment technique that employs a two-step process: the speech is first explicitly segmented (primarily into clauses), then the segments are aligned with the pen strokes. Our speech segmentation step is unique in that it uses sketch features for locating segment boundaries in multimodal dialog. In addition, it uses a single classifier to directly combine word-based, prosodic (pause), and sketch-based features. In the second step, segments are initially aligned with strokes based on temporal correlation, and then classifiers are used to detect and correct two common alignment errors. Our two-step technique has proven to be substantially more accurate at alignment than the existing technique that lacked explicit segmentation. It is more important that, for nearly all cases, our new technique results in greater gesture classification accuracy than the existing technique, and performed nearly as well as the benchmark manual speech–sketch alignment.

Classifying dialogue in high-dimensional space

The richness of multimodal dialogue makes the space of possible features required to describe it very large relative to the amount of training data. However, conventional classifier learners require large amounts of data to avoid overfitting, or do not generalize well to unseen examples. To learn dialogue classifiers using a rich feature set and fewer data points than features, we apply a recent technique, ℓ 1 -regularized logistic regression. We demonstrate this approach empirically on real data from Project LISTEN's Reading Tutor, which displays a story on a computer screen and listens to a child read aloud. We train a classifier to predict task completion (i.e., whether the student will finish reading the story) with 71% accuracy on a balanced, unseen test set. To characterize differences in the behavior of children when they choose the story they read, we likewise train and test a classifier that with 73.6% accuracy infers who chose the story based on the ensuing dialogue. Both classifiers significantly outperform baselines and reveal relevant features of the dialogue.

Learning and Evaluation of Dialogue Strategies for New Applications: Empirical Methods for Optimization from Small Data Sets

We present a new data-driven methodology for simulation-based dialogue strategy learning, which allows us to address several problems in the field of automatic optimization of dialogue strategies: learning effective dialogue strategies when no initial data or system exists, and determining a data-driven reward function. In addition, we evaluate the result with real users, and explore how results transfer between simulated and real interactions. We use Reinforcement Learning (RL) to learn multimodal dialogue strategies by interaction with a simulated environment which is “bootstrapped” from small amounts of Wizard-of-Oz (WOZ) data. This use of WOZ data allows data-driven development of optimal strategies for domains where no working prototype is available. Using simulation-based RL allows us to find optimal policies which are not (necessarily) present in the original data. Our results show that simulation-based RL significantly outperforms the average (human wizard) strategy as learned from the data by using Supervised Learning. The bootstrapped RL-based policy gains on average 50 times more reward when tested in simulation, and almost 18 times more reward when interacting with real users. Users also subjectively rate the RL-based policy on average 10% higher. We also show that results from simulated interaction do transfer to interaction with real users, and we explicitly evaluate the stability of the data-driven reward function.