The ability to infer and mediate intentions has been recognized as a crucial task in recent robotics research, where it is agreed that robots are required to be equipped with intentional mechanisms in order to participate in collaborative tasks with humans. Reasoning about - or rather, perceiving - intentions enables robots to infer what other agents are doing, to communicate what are their plans, or to take proactive decisions. Intent recognition relates to several system requirements, such as the need of an enhanced collaboration mechanism in human-machine interactions, the need for adversarial technology in competitive scenarios, ambient intelligence, or predictive security systems. When attempting to describe what an intention is, agreement exists to represent it as a plan together with the goal it attempts to achieve. Being compatible with computer science concepts, this representation enables to handle intentions with methodologies based on planning, such as the Planning Domain Description Language or Hierarchical Task Networks. In this licentiate we describe how intentions can be processed using classical planning methods, with an eye also on newer technologies such as deep networks. Our goal is to study and define computational models that would allow robotic agents to infer, construct and mediate intentions. Additionally, we explore how intentions in the form of abstract plans can be grounded to sensorial data, and in particular we provide discussion on grounding over speech utterances and affordances, that correspond to the action possibilities offered by an environment.

The goal of this thesis is to enable a robot to actively collaborate with a person to move an object in an efficient, smooth and robust manner. For a robot to actively assist a person it is key that the robot recognizes the actions or phases of a collaborative tasks. This requires the robot to have the ability to estimate a person’s movement intent. A hurdle in collaboratively moving an object is determining whether the partner is trying to rotate or translate the object (the rotation versus translation problem). In this thesis, Hidden Markov Models (HMM) are used to recognize human intent of rotation or translation in real-time. Based on this recognition, an appropriate impedance control mode is selected to assist the person. The approach is tested on a seven degree-of-freedom industrial robot, KUKA LBR iiwa 14 R820, working with a human partner during manipulation tasks. Results show the HMMs can estimate human intent with accuracy of 87.5% by using only haptic data recorded from the robot. Integrated with impedance control, the robot is able to collaborate smoothly and efficiently with a person during the manipulation tasks. The HMMs are compared with a switching function based approach that uses interaction force magnitudes to recognize rotation versus translation. The results show that HMMs can predict correctly when fast rotation or slow translation is desired, whereas the switching function based on force magnitudes performs poorly.

This thesis presents an activity mode intent recognition approach for safe, robust and reliable control of powered backbone exoskeleton. The thesis presents the background and a concept for a powered backbone exoskeleton that would work in parallel with a user. The necessary prerequisites for the thesis are presented, including the collection and processing of surface electromyography signals and inertial sensor data to recognize the user’s activity. The development of activity mode intent recognizer was described based on decision tree classification in order to leverage its computational efficiency. The intent recognizer is a high-level supervisory controller that belongs to a three-level control structure for a powered backbone exoskeleton. The recognizer uses surface electromyography and inertial signals as the input and CART (classification and regression tree) as the classifier. The experimental results indicate that the recognizer can extract the user’s intent with minimal delay. The approach achieves a low recognition error rate and a user-unperceived latency by using sliding overlapped analysis window. The approach shows great potential for future implementation on a prototype backbone exoskeleton.

Communication with customers is a major part of customer experience as well as a great source of data mining. More businesses are engaging with consumers via text messages. Before 2020, 39% of businesses already use some form of text messaging to communicate with their consumers. Many more were expected to adopt the technology after 2020[1]. Email response rates are merely 8%, compared to a response rate of 45% for text messaging[2]. A significant portion of this communication involves customer enquiries or support messages sent in both directions. According to estimates, more than 80% of today’s data is stored in an unstructured format (suchas text, image, audio, or video) [3], with a significant portion of it being stated in ambiguous natural language. When analyzing such data, qualitative data analysis techniques are usually employed. In order to facilitate the automated examination of huge corpora of textual material, researchers have turned to natural language processing techniques[4]. Under the light of shared statistics above, Billogram[5] has decided that support messages between creditors and recipients can be mined for predictive maintenance purposes, such as early identification of an outlier like a bug, defect, or wrongly built feature. As one sentence goal definition, Billogram is looking for an answer to ”why are people reaching out to begin with?” This thesis project discusses implementing unsupervised clustering of support messages by benefiting from natural language processing methods as well as performance metrics of results to answer Billogram’s question. The research also contains intent recognition of clustered messages in two different ways, one automatic and one semi-manual, the results have been discussed and compared. LDA and manual intent assignment approach of the first research has 100 topics and a 0.293 coherence score. On the other hand, the second approach produced 158 clusters with UMAP and HDBSCAN while intent recognition was automatic. Creating clusters will help identifying issues which can be subjects of increased focus, automation, or even down-prioritizing. Therefore, this research lands in the predictive maintenance[9] area. This study, which will get better over time with more iterations in the company, also contains the preliminary work for ”labeling” or ”describing”clusters and their intents.

Les récentes percées technologiques dans les domaines de l’actionnement, de la perception et de l’intelligence artificielle annoncent une nouvelle ère pour l’assistance à la conduite et les véhicules hautement automatisés. Toutefois, dans un contexte où l’automatisation demeure imparfaite, il est primordial de s’assurer que le système d’automatisation de la conduite puisse maintenir la conscience de la situation du conducteur afin que ce dernier puisse accomplir avec succès son rôle de supervision des actions du système. En même temps, le système doit pouvoir assurer la sécurité du véhicule et prévenir les actions du conducteur qui risqueraient de compromettre sa sécurité et celle des usagers de la route. Il est donc nécessaire d’intégrer dès la conception du système automatisé de conduite, la problématique des interactions avec le conducteur en réglant les problèmes de partage de tâche et de degré de liberté, d’autorité et de niveau d’automatisation du système. S’inscrivant dans le cadre du projet ANR-CoCoVeA (Coopération Conducteur-Véhicule Automatisé), cette thèse se penche de plus près sur la question de la coopération entre l’automate de conduite et le conducteur. Notre objectif est de fournir au conducteur un niveau d’assistance conforme à ses attentes, capable de prendre en compte ses intentions tout en assurant un niveau de sécurité important. Pour cela nous proposons un cadre général qui intègre l’ensemble des fonctionnalités nécessaires sous la forme d’une architecture permettant une coopération à plusieurs niveaux de la tâche de conduite. Les notions d’attribution des tâches et de gestion d’autorité avec leurs différentes nuances sont abordées et l’ensemble des fonctions du système identifiées dans l’architecture ont été étudiées et adaptées pour ce besoin de coopération. Ainsi, nous avons développé des algorithmes de décision de la manœuvre à effectuer, de planification de trajectoire et de contrôle qui intègrent des mécanismes leur permettant de s’adapter aux actions et aux intentions du conducteur lors d’un éventuel conflit. En complément de l’aspect technique, cette thèse étudie les notions de coopération sous l’angle des facteurs humains en intégrant des tests utilisateur réalisés sur le simulateur de conduite dynamique SHERPA-LAMIH. Ces tests ont permis à la fois de valider les développements réalisés et d’approfondir l’étude grâce à l’éclairage qu’ils ont apporté sur l’intérêt de chaque forme de coopération. / The recent technological breakthroughs in the actuation, perception and artificial intelligence domains herald a new dawn for driving assistance and highly automated driving. However, in a context where the automation remains imperfect and prone to error, it is crucial to ensure that the automated driving system maintains the driver’s situation awareness in order to be able to successfully and continuously supervise the system’s actions. At the same time, the system must be able to ensure the safety of the vehicle and prevent the driver’s actions that would compromise his safety and that of other road users. Therefore, it is essential that the issue of interaction and cooperation with the driver is addressed throughout the whole system design process. This entails the issues of task allocation, authority management and levels of automation. Conducted in the scope of the projet ANR-CoCoVeA (French acronym for: "Cooperation between Driver and Automated Vehicle"), this thesis takes a closer look at the question of cooperation between the driver and automated driving systems. Our main objective is to provide the driver with a suitable assistance level that accounts for his intentions while ensuring global safety. For this matter, we propose a general framework that incorporates the necessary features for a successful cooperation at the different levels of the driving task in the form of a system architecture. The questions of task allocation and authority management are addressed under their different nuances and the identified system functionalities are studied and adapted to match the cooperation requirements. Therefore, we have developed algorithms to perform maneuver decision making, trajectory planning, and control that include the necessary mechanisms to adapt to the driver’s actions and intentions in the case of potential conflicts. In addition to the technical aspects, this thesis studies the cooperation notions from the human factor perspective. User test studies conducted on the SHERPA-LAMIH dynamic simulator allowed for the validation of the different developments while shedding light on the benefits of different cooperation forms.