Questioning Wizard of Oz: Effects of Revealing the Wizard behind the Robot

Wizard of Oz, a very commonly employed technique in human-robot interaction, faces the criticism of being deceptive as the humans interacting with the robot are told, if at all, only at the end of their interaction that there was in fact a human behind the robot. What if the robot reveals the wizard behind itself very early in the interaction? We built a deep wizard of Oz setup to allow for a robot to play together with a human against a computer AI in the context of Connect 4 game. This cooperative game interaction against a common opponent is then followed by a conversation between the human and the robot. We conducted an exploratory user study with 29 adults with three conditions where the robot reveals the wizard, lies about the wizard, and does not say anything, respectively. We also split the data based on how the participants perceive the robot in terms of autonomy. Using different metrics, we evaluate how the users interact with and perceive the robot in both the experimental and perceived conditions. We find that while there is indeed a significant difference in the participants willingness to follow robots suggestions between the experimental conditions as well as in the effort they put to prove themselves as humans (reverse Turing test), there isn’t any significant difference in their robot perception. Additionally, how humans perceive whether the robot is tele-operated or autonomous seems to be indifferent to the robot revealing its identity, i.e., the pre-conceived notions may be uninfluenced even if the robot explicitly states otherwise. Lastly, interestingly in the perception based conditions, absence of statistical significance may suggest that, in certain contexts, wizard of oz may not require hiding the wizard after all.

Error handling in multimodal voice-enabled interfaces of tour-guide robots using graphical models

Plamen Prodanov

Mobile service robots are going to play an increasing role in the society of humans. Voice-enabled interaction with service robots becomes very important, if such robots are to be deployed in real-world environments and accepted by the vast majority of potential human users. The research presented in this thesis addresses the problem of speech recognition integration in an interactive voice-enabled interface of a service robot, in particular a tour-guide robot. The task of a tour-guide robot is to engage visitors to mass exhibitions (users) in dialogue providing the services it is designed for (e.g. exhibit presentations) within a limited time. In managing tour-guide dialogues, extracting the user goal (intention) for requesting a particular service at each dialogue state is the key issue. In mass exhibition conditions speech recognition errors are inevitable because of noisy speech and uncooperative users of robots with no prior experience in robotics. They can jeopardize the user goal identification. Wrongly identified user goals can lead to communication failures. Therefore, to reduce the risk of such failures, methods for detecting and compensating for communication failures in human-robot dialogue are needed. During the short-term interaction with visitors, the interpretation of the user goal at each dialogue state can be improved by combining speech recognition in the speech modality with information from other available robot modalities. The methods presented in this thesis exploit probabilistic models for fusing information from speech and auxiliary modalities of the robot for user goal identification and communication failure detection. To compensate for the detected communication failures we investigate multimodal methods for recovery from communication failures. To model the process of modality fusion, taking into account the uncertainties in the information extracted from each input modality during human-robot interaction, we use the probabilistic framework of Bayesian networks. Bayesian networks are graphical models that represent a joint probability function over a set of random variables. They are used to model the dependencies among variables associated with the user goals, modality related events (e.g. the event of user presence that is inferred from the laser scanner modality of the robot), and observed modality features providing evidence in favor of these modality events. Bayesian networks are used to calculate posterior probabilities over the possible user goals at each dialogue state. These probabilities serve as a base in deciding if the user goal is valid, i.e. if it can be mapped into a tour-guide service (e.g. exhibit presentation) or is undefined – signaling a possible communication failure. The Bayesian network can be also used to elicit probabilities over the modality events revealing information about the possible cause for a communication failure. Introducing new user goal aspects (e.g. new modality events and related features) that provide auxiliary information for detecting communication failures makes the design process cumbersome, calling for a systematic approach in the Bayesian network modelling. Generally, introducing new variables for user goal identification in the Bayesian networks can lead to complex and computationally expensive models. In order to make the design process more systematic and modular, we adapt principles from the theory of grounding in human communication. When people communicate, they resolve understanding problems in a collaborative joint effort of providing evidence of common shared knowledge (grounding). We use Bayesian network topologies, tailored to limited computational resources, to model a state-based grounding model fusing information from three different input modalities (laser, video and speech) to infer possible grounding states. These grounding states are associated with modality events showing if the user is present in range for communication, if the user is attending to the interaction, whether the speech modality is reliable, and if the user goal is valid. The state-based grounding model is used to compute probabilities that intermediary grounding states have been reached. This serves as a base for detecting if the the user has reached the final grounding state, or wether a repair dialogue sequence is needed. In the case of a repair dialogue sequence, the tour-guide robot can exploit the multiple available modalities along with speech. For example, if the user has failed to reach the grounding state related to her/his presence in range for communication, the robot can use its move modality to search and attract the attention of the visitors. In the case when speech recognition is detected to be unreliable, the robot can offer the alternative use of the buttons modality in the repair sequence. Given the probability of each grounding state, and the dialogue sequence that can be executed in the next dialogue state, a tour-guide robot has different preferences on the possible dialogue continuation. If the possible dialogue sequences at each dialogue state are defined as actions, the introduced principle of maximum expected utility (MEU) provides an explicit way of action selection, based on the action utility, given the evidence about the user goal at each dialogue state. Decision networks, constructed as graphical models based on Bayesian networks are proposed to perform MEU-based decisions, incorporating the utility of the actions to be chosen at each dialogue state by the tour-guide robot. These action utilities are defined taking into account the tour-guide task requirements. The proposed graphical models for user goal identification and dialogue error handling in human-robot dialogue are evaluated in experiments with multimodal data. These data were collected during the operation of the tour-guide robot RoboX at the Autonomous System Lab of EPFL and at the Swiss National Exhibition in 2002 (Expo.02). The evaluation experiments use component and system level metrics for technical (objective) and user-based (subjective) evaluation. On the component level, the technical evaluation is done by calculating accuracies, as objective measures of the performance of the grounding model, and the resulting performance of the user goal identification in dialogue. The benefit of the proposed error handling framework is demonstrated comparing the accuracy of a baseline interactive system, employing only speech recognition for user goal identification, and a system equipped with multimodal grounding models for error handling.

EPFL2006

Motion Learning and Adaptive Impedance for Robot Control during Physical Interaction with Humans

Aude Billard, Elena Gribovskaya

One of the hallmarks of physical interaction between humans is haptic communication, i.e. an informa- tion exchange through force signals. Humans excel in tasks that require such interaction by adapting impedance and anticipating the partner’s intentions. It is highly desirable to endow robots with similar capabilities. Recently, the robotics community renewed its interest in variable impedance control. A special emphasis is put on the development of controllers that incorporate learning as an essential element. This article combines programming by demonstration and adaptive control for teaching a robot to physically interact with a human in a collaborative task requiring sharing of a load by the two partners. Learning a task model allows the robot to anticipate the partner’s intentions and adapt its motion according to perceived forces. As the human represents a highly complex contact environment, direct reproduction of the learned model may lead to sub-optimal results. To compensate for unmodelled uncertainties, in addition to learning we propose an adaptive control algorithm which tunes the impedance parameters, so as to ensure accurate reproduction. To simplify the illustration of the concepts introduced in this paper and provide a systematic evaluation, we present experimental results obtained in physically-realistic simulation of a dyad of two planar 2-DOF robots.

2011

From human-intention recognition to compliant control using dynamical systems in physical human-robot interaction

Mahdi Khoramshahi

Human ability to coordinate one's actions with other individuals to perform a task together is fascinating. For example, we coordinate our action with others when we carry a heavy object or when we construct a piece of furniture. Capabilities such as (1) force/compliance adaptation, (2) intention recognition, and (3) action/motion prediction enables us to assist others and fulfill the task. For instance, by adapting the compliance, we not only reject undesirable perturbations that undermine the task but also incorporate others' motions into the interaction. Complying with partners' motions allows us to recognize their intention and consequently predict their actions. With the growth of factories involving humans and robots working side by side, designing controllers and algorithms with such capacities is a crucial step toward assistive robotics. The challenge, however, is to attain a unified control strategy with predictive/adaptive capacities at the task, motion, and force-level which ensures a stable and safe interaction. To this aim, this thesis proposes a state-dependent dynamical system-based approach for prediction and control in physical human-robot interactions. In the first part of this dissertation, we focus on the human capacity to predict their partners' motion. More specifically, we investigate mechanisms of spatio-temporal coordination between two partners. We employ a simple scenario called âthe mirror gameâ where two individuals (human, robot, or avatar) imitate each other's motions. Our empirical assessment reveal that the intention-based prediction of the leader's motions allows the follower to compensate for perception-action delays and to improve the tracking performance in terms of temporal coordination and confidence. In the second part of this dissertation, we propose an adaptive mechanism that enables the robot to recognize the intention of a human user. We utilize state-dependent dynamical systems for motion planning and impedance control to deliver safe and compliant human-robot interaction. We consider a series of tasks (possible human intentions) encoded by dynamical system. Applying a similarity metric between the real velocities (induced by the human) and desired velocities generated by the dynamical systems, the robot is thus able to recognize the human's intention and switch to the intended task. We provide a rigorous experimental and analytical evaluation of our method yielding an interaction behavior that is safe and intuitive for the human. Finally, we tackle the compliance adaptation capability. We propose an admittance controller that reacts only when human-intentional forces are detected. Intentional and accidental forces are distinguished by measuring the persistency of the external forces, through a computation of the autocorrelation/energy of the force patterns. The overall controller exhibits variable stiffness where high stiffness allows the robot to reject the external disturbances and execute the task autonomously whereas low stiffness enables the robot to comply with human intentional forces. We demonstrate that our control architecture is effective in delivering satisfactory tracking and compliant behavior through a series of robotic experimentations.