Search

Despite the fact that task-oriented conversation systems have received much attention from the dialogue research community, only a handful of them have been studied in a real-world manufacturing context using industrial robots. One stumbling block is the lack of a domain-specific discourse corpus for training these systems. Another difficulty is that earlier attempts to integrate natural language interfaces (such as chatbots) into the industrial sector have primarily focused on task completion rates. When designing a dialogue system for social robots, the user experience is prioritized above industrial robots. To overcome these challenges, we provide the Industrial Robots Domain Wizard-of-Oz dataset (IRWoZ), a fully-labeled discourse dataset covering four robotics domains. It delivers simulated discussions between shop floor workers and industrial robots, with over 401 dialogues, to promote language-assisted Human-Robot Interaction (HRI) in industrial settings. Small talk concepts and human-to-human conversation strategies are provided to support human-like answer generation and give a more natural and adaptable dialogue environment to increase user experience and engagement. Finally, we propose and evaluate an end-to-end Task-oriented Dialogue for Industrial Robots (ToD4IR) using two types of pre-trained backbone models: GPT-2 and GPT-Neo, on the IRWoZ dataset. We performed a series of trials to validate ToD4IR's performance in a real manufacturing context. Our experiments demonstrate that ToD4IR outperforms three downstream task-oriented dialogue tasks, i.e., dialogue state tracking, dialogue act generation, and response generation, on the IRWoZ dataset. Our source code of ToD4IR and the IRWoZ dataset is accessible at https://github.com/lcroy/ToD4IR for reproducible research.

International audience; In this article, we consider the problem of delivering an object to a human coworker by means of an aerial robot (AR). To this aim, we present an ergonomics-aware Nonlinear Model Predictive Control (NMPC) designed to autonomously perform the handover. The method is general enough to be applied to any multi-rotor aerial vehicle (MRAV) with a minimal adaptation of the robot model. The formulation of the optimal control problem steers the AR toward a handover location by optimizing the human coworker ergonomics, which includes the predicted arm joint torques of the human. The motion task is expressed in a frame relative to the human, whose motion model is included in the equations of the NMPC. This allows the controller to promptly adapt to the human movements by predicting her future poses over the horizon. The control framework also accounts for the problem of maintaining visibility on the human coworker, while respecting both the actuation and state limits of the robot. Additionally, a safety barrier is embedded in the controller to avoid any risk of collision with the human partner. Realistic simulations are performed to validate the feasibility of the approach and the source code of the implementation is released open-source.

Interpreting the behaviour of autonomous machines will be a daily activity for future generations. Yet, surprisingly little is currently known about how people ascribe intentions to human-like and non-human-like agents or objects. In a series of six experiments, we compared people’s ability to extract non-mentalistic (i.e., where an agent is looking) and mentalistic (i.e., what an agent is looking at; what an agent is going to do) information from identical gaze and head movements performed by humans, human-like robots, and a non-human-like object. Results showed that people are faster to infer the mental content of human agents compared to robotic agents. Furthermore, the form of the non-human entity may differently engage mentalizing processes depending on how human-like its appearance is. These results are not easily explained by non-mentalizing strategies (e.g., spatial accounts), as we observed no clear differences in control conditions across the three different agents. Overall, results suggest that human-like robotic actions may be processed differently from both humans’ and objects’ behaviour. We discuss the extent to which these findings inform our understanding of the relevance of an agents’ or objects’ physical features in triggering mentalizing abilities and its relevance for human–robot interaction.

This is a post-peer-review, pre-copyedit version of an article published in Applied Technologies. The final authenticated version is available online at: http://dx.doi.org/10.1007/978-3-030-42520-3_42 This paper discusses about elements to be considered for developing a Service Robot that performs its task in a social environment. Due to the social focus of the service, not only technical considerations are demanded in order to accomplish with the task, but also the acceptance of use for the people, who interact with all of them. As our particular research topic, we establish a taxonomy to determine the framework for the development of socially-aware robot assistants for serving tasks such as deliveries. This is a general approach to be considered for any service robot being implemented in a social context. This article presents several previous cases of the implementation of service mobile robots, their analysis and the motivation of how to solve their acceptance and use by people. Therefore, under this approach it is very important not to generate false expectations about the capabilities of the robot, because as it is explained in the state of the art analysis that very high unsatisfied expectations lead to leaving the robot unused....

Humanoids have increasingly become the focus of attention in robotic research, especially in rehabilitation robotics. In the past few years, the prevalence of Cerebral Palsy (CP) in the world population has remained unchanged. CP is one of the most common disorders affecting muscle control and coordination in children, resulting in movement disorders. There are assistive technologies that are used for increasing, maintaining or improving the functional abilities of CP children. However, there is still a lack of interventional; therapy which involves Social Assistive Robotics (SAR). Socially assistive technologies have great potential in education & learning for children, especially for children with disabilities. Problems faced by children with CP may not be confined to physical impairment but may in addition have impairments of cognition, visual, language and communication. Therefore, combining physical, speech and language and occupational therapy with cognitive and social development is very important. This paper aims to describe the architecture of using humanoid robot NAO as a tool to augment physical learning for CP children. This work is mainly based on qualitative study, which is aimed to analyze how the HRI impacts the motor gross function and attention. This paper will describe the aim, objectives, methodology, and experiment layout. Experiment with CP children will be done in the near future after obtaining ethical approval from UiTM Ethics Committee. The outcome of the study may exhibit positive utilization of the humanoid robot NAO as a new intervention tool to augment therapy for children with CP.

International audience; For intelligent Human-Robot Interaction (HRI), a robot should be equipped with some core reasoning capabilities such as perspective taking, effort analysis, and affordance analysis. This paper starts to explore how a robot equipped with such reasoning abilities could be evaluated. To this end, inspired by the Turing test, we design a game involving a human-robot competition scenario. Interestingly, the participants' subjective feedback, which tended to compare the robot's abilities with their own, points toward potential criteria for developing benchmark scenarios and evaluation matrices.

This paper presents an experimental evaluation of physical human-human interaction in lightweight condition using a one degree of freedom robotized setup. It explores possible origins of Physical Human-Human communication, more precisely, the hypothesis of a time based communication. To explore if the communication is correlated to time a statistical state machine model based on physical Human-Human interaction is proposed. The model is tested with 14 subjects and presents results that are close to human-human performances.

Multimodal user interface is a kind of user interface which is able to receive various high data and produce various types of output in reaction to it. Though human – robot interaction is based on human and computer interaction, it is more complex than human – computer interaction due to considerable increase in the number of inputs and outputs. Increase in parameters results from mobility of robot and human, interaction of robot and objects and necessity for three-dimensional understanding of the environment. Security is one of the main concerns in human – robot interaction. One of the novel applications of robots is caring for people with physical disability and aged people. Due to some limitations of these patients, simplicity of communication and no need for learning by the clients, in one hand, and presence of various inputs, versatility and automatic adaptation of robot performance and ability to respond to incomplete input, on the other hand, is palpable.

International audience; In this paper we present a system for automatic prediction of extraversion during the first thin slices of human-robot interaction (HRI). This work is based on the hypothesis that personality traits and attitude towards robot appear in the behavioural response of humans during HRI. We propose a set of four non-verbal movement features that characterize human behavior during interaction. We focus our study on predicting Extraversion using these features , extracted from a dataset consisting of 39 healthy adults interacting with the humanoid iCub. Our analysis shows that it is possible to predict to a good level (64%) the Extraversion of a human from a thin slice of interaction relying only on non-verbal movement features. Our results are comparable to the state-of-the-art obtained in HHI [ 23 ] .

Combining multiple scientific disciplines, robotic technology has made significant progress the last decade, and so did the interactions between humans and robots. This article updates the agenda for robotic research by highlighting thefactors that affect Human – Robot Interaction (HRI), and explains the relationships and dependencies that exist between them. The four main factors that define the properties of a robot, and therefore the interaction, are distributed in two dimensions: (1) Intelligence (Control - Autonomy), and (2) Perspective (Tool -Medium). Based on these factors, we introduce a generic model for comparing and contrasting robots (CCM), aiming to provide a common platform for robot designers, developers and users. The framework for HRI we propose stems mainly from the vagueness and the lack of clarity that has been observed in the definitions of both Direct and Indirect HRI.