Designing an effective human-robot interaction paradigm is particularly important for complex tasks such as multi-robot manipulation that require the human and robot to work together in a tightly coupled fashion. Although increasing the number of robots can expand the area that the robots can cover within a bounded period of time, a poor human-robot interface will ultimately compromise the performance of the team of robots. However, introducing a human operator to the team of robots, does not automatically improve performance due to the difficulty of teleoperating mobile robots with manipulators. The human operator’s concentration is divided not only among multiple robots but also between controlling each robot’s base and arm. This complexity substantially increases the potential neglect time, since the operator’s inability to effectively attend to each robot during a critical phase of the task leads to a significant degradation in task performance. There are several proven paradigms for increasing the efficacy of human-robot interaction: 1) multimodal interfaces in which the user controls the robots using voice and gesture; 2) configurable interfaces which allow the user to create new commands by demonstrating them; 3) adaptive interfaces which reduce the operator’s workload as necessary through increasing robot autonomy. This dissertation presents an evaluation of the relative benefits of different types of user interfaces for multi-robot systems composed of robots with wheeled bases and three degree of freedom arms. It describes a design for constructing low-cost multi-robot manipulation systems from off the shelf parts. User expertise was measured along three axes (navigation, manipulation, and coordination), and participants who performed above threshold on two out of three dimensions on a calibration task were rated as expert. Our experiments reveal that the relative expertise of the user was the key determinant of the best performing interface paradigm for that user, indicating that good user modiii eling is essential for designing a human-robot interaction system that will be used for an extended period of time. The contributions of the dissertation include: 1) a model for detecting operator distraction from robot motion trajectories; 2) adjustable autonomy paradigms for reducing operator workload; 3) a method for creating coordinated multi-robot behaviors from demonstrations with a single robot; 4) a user modeling approach for identifying expert-novice differences from short teleoperation traces.

No description available.

Cette thèse présente la conception d'une interface haptique capable de rendre l'interaction physique humain-robot naturelle et intuitive. Il s'agit là d'un sujet d'étude très important avec l'avènement de la robotique collaborative et la présence toujours accrue des robots dans la vie de tous les jours. Les travaux présentés se concentrent sur l'approche macro-mini à titre d'interface haptique, plus particulièrement trois aspects importants lors de la conception d'un système macro-mini. Le premier chapitre permet d'apprendre à parler physiquement au robot (lui faire comprendre les intentions de l'humain) dans un contexte de déplacement collaboratif. Plus particulièrement, il consiste à comparer différentes méthodes pour traduire les déplacements (ou les intentions) de l'humain à un robot. Dans ce cas, des coquilles à faible impédance sont attachées sur les membrures d'un manipulateur sériel. L'humain interagit avec le robot en déplaçant ces coquilles. La réponse du robot est alors de se déplacer de façon à ce que ses membrures suivent les déplacements de la coquille qui leur est associée pour ainsi les conserver dans leur configuration neutre. Le déplacement d'une coquille par rapport à sa membrure est considéré comme une vitesse désirée de ladite membrure. Il s'agit donc de résoudre le problème cinématique inverse pour traduire le déplacement de la coquille en déplacement articulaire. Cependant, différentes stratégies peuvent être employées pour résoudre ce problème. Ce projet vise donc à comparer l'efficacité de ces méthodes. Pour y parvenir, une étude générale de ces méthodes est réalisée. Puis, un formalisme mathématique est décrit pour adapter ces méthodes à l'application présente. En effet, en fonction du type de coquille et de la membrure, tous les degrés de liberté ne sont pas nécessairement possibles. Ce formalisme mathématique permet de tenir compte de ces contraintes. Ensuite, des simulations sont réalisées pour observer le comportement des méthodes étudiées et un indice de performance est choisi pour les comparer. Ensuite, une fois que le robot est en mesure de comprendre efficacement les intentions humaines, le problème de conception consiste à déterminer comment détecter ses intentions à l'aide d'une interface et surtout, la taille que cette interface doit prendre pour bien parler. En d'autres mots, le second chapitre présente une analyse de l'impact du débattement d'un mécanisme mini actif sur la bande passante mécanique de mouvements possibles lors de la manipulation de charges lourdes. En effet, l'approche macro-mini utilise généralement un robot mini passif, ce qui fait que l'utilisateur ressent toute l'inertie de la charge. Lorsque la charge devient suffisamment lourde, il est nécessaire pour le mini d'appuyer l'utilisateur en fournissant une force pour conserver l'interaction naturelle. Ceci signifie que le mini doit être actionné, i.e., actif. Il est cependant important que le mini reste rétrocommandable pour le bon fonctionnement de l'approche macro-mini. Des modèles mathématiques du système sont donc présentés. Les contraintes relatives à l'application sont décrites ainsi que leur impact sur la bande passante. À l'aide d'un contrôleur simple, des simulations sont réalisées à l'aide des outils développés pour déterminer le débattement nécessaire du mini actif qui permet la bande passante désirée. Enfin, une interface haptique capable de reproduire une poignée de main naturelle et intuitive avec un robot est présentée. Ce chapitre peut être divisé en deux aspects, i.e., la main et le bras. Ici, la main est le robot mini et le bras, le robot macro. D'abord, un prototype de main robotique est conçu et fabriqué. Inspirée de l'anatomie humaine, cette main robotique possède une paume comprimable capable d'émuler celle de l'humain ainsi que trois doigts sous-actionnés. Un pouce passif, relié au niveau de compression de la paume, complète le tout. Le contrôle de la main se fait via une position avec rétroaction, et ce, pour chacun des deux actionneurs (un pour la paume, l'autre pour les trois doigts). Ensuite, la main robotique est montée sur un manipulateur sériel collaboratif (le Kuka LWR), le bras. Ce dernier est contrôlé en impédance autour d'une trajectoire harmonique dans un plan vertical. En fonction des paramètres de la trajectoire (amplitude, fréquence, coefficients d'amortissement et de raideur), ce prototype permet de conférer une personnalité active au robot. L'expérimentation faite auprès de sujets humains permet de déterminer les valeurs considérées plus naturelles pour les différents paramètres de la trajectoire ainsi que diverses pistes à explorer pour des travaux futurs. / This thesis presents the design of a haptic interface capable of rendering a physical human-robot interaction natural and intuitive. It is a very important subject to study with the rise in collaborative robots and the ever-increasing presence of robots in everyday life. The work presented here focuses on the macro-mini architecture as haptic interface, more precisely on three important aspects to consider during the design of a macro-mini system. The first chapter explores how to physically communicate with a robot (make it understand the human's intentions) in a context of collaborative motion. In more details, the goal is to compare different methods to translate human motions (or intentions) to a robot. In this case, low impedance passive articulated shells are mounted on the links of a serial manipulator. The human operator interacts with the robot by displacing the shells. The robot's response is then to move so that its links follow the motion of their associated shell. The better the robot can follow the shells, the closest to their neutral configuration the shells can remain. The shell displacement relative to its link is considered as a desired velocity of the link. The translation of the shell displacement into joint motion then becomes an inverse kinematic problem. Different strategies can be used to solve this problem. This project then aims at comparing the efficiency of those strategies. To this end, a general study of the different strategies is performed. Then, a mathematical formalism is described, adapting said strategies to the present context. Indeed, depending on the type of shell and the position of the link in the chain, all degrees of freedom are not necessarily possible. This formalism takes these limitations into account. Then, simulations are conducted to observe the behaviour of the different strategies studied and a performance index is chosen to compare them. Afterwards, once the robot is capable of efficiently understanding the human intentions, the next step is to determine how to detect the intentions with the help of an interface and, most of all, what size should this interface have so as to communicate well. In other words, the second chapter presents an analysis of the impact that the range of motion of an active mini mechanism has on the mechanical bandwidth for the possible motions during the handling of large payloads. Indeed, the macro-mini architecture generally uses a passive mini robot, which means that the human operator feels the whole inertia of the payload. When the payload becomes sufficiently heavy, it becomes necessary for the mini robot to help the operator by working as well so as to keep the interaction natural. This means that the mini robot should then be actuated, i.e., active. It is however important that the mini robot remains backdrivable for the macro-mini architecture to work properly. Mathematical models are then presented. The limitations related to the application are described, as well as their effect on the bandwidth. With the help of a simple controller, simulations are performed with the tools developed to determine the range of motion necessary for the active mini robot which would allow the desired bandwidth. Finally, a haptic interface capable of emulating a natural and intuitive handshake with a robot is presented. This chapter can be divided into two aspects, i.e., the hand and the arm. Here, the hand is the mini robot and the arm, the macro robot. First, a robotic hand prototype is designed and constructed. Inspired by the human anatomy, this robotic hand has a compliant palm able to emulate a human palm as well as three underactuated fingers. A passive thumb, tied to the palm compression level, completes the hand. The hand control is done with position control with feedback for both actuators (one for the palm, the other for the three fingers). Then, the robotic hand is mounted on a collaborative serial manipulator (the Kuka LWR), the arm. The arm is controlled in impedance around a harmonic trajectory in a vertical plane. Depending on the parameters for the trajectory (amplitude, frequency, stiffness and damping coefficients), this prototype provides an active personality to the robot. Experimentation is conducted with human subjects to determine the values considered more natural for the different trajectory parameters as well as several improvements for the prototype in future works.

Effective locomotion and maneuvering in aquatic environments is important for survival for marine fauna. The ability to move quickly, change direction, and tune energy consumption for long migrations is critical for escape from predators and pursuit of prey. This controlled propulsion in terms of varying speed, turning rates, and actuation effort is of interest for the next generation of underwater vehicle design. Integration of biological functional simplicity, robustness, and superior performance enables robotic vehicles to successfully complete difficult and dynamic operational goals. Gelatinous animals known as Cnidarians employ a wide variety of propulsive methods, ranging from the simple but efficient propulsion of large jellyfish to the rapid and highly maneuverable multi-jet propulsion of colonial animals known as siphonophores. This dissertation studies how these two extremes of underwater soft body propulsion are able to achieve simple yet effective control and locomotion, and thus inform the design of effective vehicle propulsion control and actuation. Two large single bell jellyfish robots, Cyro 2 and Cyro 3, were designed and constructed to implement the simple body form and propulsive methods of large jellyfish to study the unique locomotive characteristics and fluid interactions that generate straight swimming and turning maneuvers. The other extreme of small soft-body colonies moving by multi-jet propulsion was subsequently investigated in-depth, starting with a characterization of the biological fluid jetting actions and gaits. The results of these performance capabilities were then applied to an experimental robotic model with bio-inspired construction and controls to verify an elegant but highly functional neurological control scheme and the kinematic capabilities from varying jetting gait patterns. / PHD / The ability to move rapidly in any direction is a primary characteristic of successful animal species. Evasion of predators, as well as pursuit of prey, is paramount for survival. Jellyfish are excellent examples of animals that have thrived for millions of years with varied methods of moving in their diverse environments. However, the propulsion methods of large jellyfish in straight swimming and turning have not been well understood until recent years. This dissertation focuses on the fundamental understanding of the locomotion and fluid interaction that jellyfish use for propulsion. A large jellyfish robot, named Cyro 2 (“Cy” for the species Cyanea, “ro” for robot, and the second generation of the design), was constructed to explore the role of various structural and fluidic parameters on the locomotion characteristics of the largest jellyfish species Cyanea. The successor Cyro 3 was designed to mimic the complex motions of large jellyfish during maneuvering. Motion tracking and fluid analysis of the robot during turning was utilized to explain how jellyfish dynamically control their orientation. These results inspired further study of a unique relative of jellyfish, the siphonophore, which can swim with a modular chain of soft pumping bodies that coordinate without a central nervous system. This unique control strategy and method of movement underwater was studied by analyzing specimens of the siphonophore Nanomia. Development and modeling of elegant control techniques inspired by this species is presented and implemented on an experimental model that uses this unique propulsion method to validate and expand upon observations of live specimens. Combined, the results obtained in this dissertation open the possibility of designing advanced underwater vehicles.

Telemanipulators allow a user to interact with a potentially dangerous environment remotely. Deploying a robot arm from a UAV would allow an operator to reach farther and quicker than he or she would with a ground robotics system alone. This thesis will discuss the design and fabrication of a compact, light, 3 degree of freedom robot arm using common off the shelf products and machined components that in combination can pick up half pound samples and has a reach of 260 mm. Also addressed is making the telemanipulator interface easier to use. One of the challenges in using a robot arm with a single camera in a beyond line-of-sight scenario is the difficulty of interacting with the environment because of a loss of depth information. This lack of information can be remedied with additional sensors. Once depth to an object of interest is known, the sampler can automatically pick up objects of interest. The manipulator arm will be used in conjunction with systems developed by the Unmanned Systems Laboratory at Virginia Tech. This group is developing a unmanned ground vehicle to be carried in the payload pod of a unmanned aerial vehicle. The robot's ultimate objective is to collect shrapnel and bomb material from potentially dangerous environments. / Master of Science

This dissertation addresses the challenge of robot decision making in the presence of uncertainty, specifically focusing on robot motion decisions in the context of deep learning-based perception uncertainty. The first part of this dissertation introduces a risk-aware framework for path planning and assignment of multiple robots and multiple demands in unknown environments. The second part introduces a risk-aware motion model for searching for a target object in an unknown environment. To illustrate practical application, consider a situation such as disaster response or search-and-rescue, where it is imperative for ground vehicles to swiftly reach critical locations. Afterward, an agent deployed at a specified location must navigate inside a building to find a target, whether it is an object or a person. In the first problem, the terrain information is only available as an aerial georeferenced image frame. Semantic segmentation of the aerial images is performed using Bayesian deep learning techniques, creating a cost map for the safe navigation ground robots. The proposed framework also accounts for risk at a further level, using conditional value at risk (CVaR), for making risk-aware assignments between the source and goal. When the robot reaches its destination, the second problem addresses the object search task using a proposed machine learning-based intelligent motion model. A comparison of various motion models, including a simple greedy baseline, indicates that the proposed model yields more risk-aware and robust results. All in all, considering uncertainty in both systems leads to demonstrably safer decisions. / Doctor of Philosophy / Scientists need to demonstrate that robots are safe and reliable outside of controlled lab environments for real-world applications to be viable. This dissertation addresses the challenge of robot decision-making in the face of uncertainty, specifically focusing on robot motion decisions in the context of deep learning-based perception uncertainty. Deep learning (DL) refers to using large hierarchical structures, often called neural networks, to approximate semantic information from input data. The first part of this dissertation introduces a risk-aware framework for path planning and assignment of multiple robots and multiple demands in unknown environments. Path planning involves finding a route from the source to the goal, while assignment focuses on selecting source-goal paths to fulfill all demands. The second part introduces a risk-aware motion model for searching for a target object in an unknown environment. Being risk-aware in both cases means taking uncertainty into account. To illustrate practical application, consider a situation such as disaster response or search-and-rescue, where it is imperative for ground vehicles to swiftly reach critical locations. Afterward, an agent deployed at a specified location must navigate inside a building to find a target, whether it is an object or a person. In this dissertation, deep learning is used to interpret image inputs for two distinct robot systems. The input to the first system is an aerial georeferenced image; the second is an indoor scene. After the images are interpreted by deep learning, they undergo further processing to extract information about uncertainty. The information about the image and the uncertainty is used for later processing. In the first case, we use both a traditional path planning method and a novel path assignment method to assign one path from each source to a demand location. In the second case, a motion model is developed using image data, uncertainty, and position in relation to the anticipated target. Several potential motion models are compared for analysis. All in all, considering uncertainty in both systems leads to demonstrably safer decisions.

Cette thèse présente la conception d'une interface haptique capable de rendre l'interaction physique humain-robot naturelle et intuitive. Il s'agit là d'un sujet d'étude très important avec l'avènement de la robotique collaborative et la présence toujours accrue des robots dans la vie de tous les jours. Les travaux présentés se concentrent sur l'approche macro-mini à titre d'interface haptique, plus particulièrement trois aspects importants lors de la conception d'un système macro-mini. Le premier chapitre permet d'apprendre à parler physiquement au robot (lui faire comprendre les intentions de l'humain) dans un contexte de déplacement collaboratif. Plus particulièrement, il consiste à comparer différentes méthodes pour traduire les déplacements (ou les intentions) de l'humain à un robot. Dans ce cas, des coquilles à faible impédance sont attachées sur les membrures d'un manipulateur sériel. L'humain interagit avec le robot en déplaçant ces coquilles. La réponse du robot est alors de se déplacer de façon à ce que ses membrures suivent les déplacements de la coquille qui leur est associée pour ainsi les conserver dans leur configuration neutre. Le déplacement d'une coquille par rapport à sa membrure est considéré comme une vitesse désirée de ladite membrure. Il s'agit donc de résoudre le problème cinématique inverse pour traduire le déplacement de la coquille en déplacement articulaire. Cependant, différentes stratégies peuvent être employées pour résoudre ce problème. Ce projet vise donc à comparer l'efficacité de ces méthodes. Pour y parvenir, une étude générale de ces méthodes est réalisée. Puis, un formalisme mathématique est décrit pour adapter ces méthodes à l'application présente. En effet, en fonction du type de coquille et de la membrure, tous les degrés de liberté ne sont pas nécessairement possibles. Ce formalisme mathématique permet de tenir compte de ces contraintes. Ensuite, des simulations sont réalisées pour observer le comportement des méthodes étudiées et un indice de performance est choisi pour les comparer. Ensuite, une fois que le robot est en mesure de comprendre efficacement les intentions humaines, le problème de conception consiste à déterminer comment détecter ses intentions à l'aide d'une interface et surtout, la taille que cette interface doit prendre pour bien parler. En d'autres mots, le second chapitre présente une analyse de l'impact du débattement d'un mécanisme mini actif sur la bande passante mécanique de mouvements possibles lors de la manipulation de charges lourdes. En effet, l'approche macro-mini utilise généralement un robot mini passif, ce qui fait que l'utilisateur ressent toute l'inertie de la charge. Lorsque la charge devient suffisamment lourde, il est nécessaire pour le mini d'appuyer l'utilisateur en fournissant une force pour conserver l'interaction naturelle. Ceci signifie que le mini doit être actionné, i.e., actif. Il est cependant important que le mini reste rétrocommandable pour le bon fonctionnement de l'approche macro-mini. Des modèles mathématiques du système sont donc présentés. Les contraintes relatives à l'application sont décrites ainsi que leur impact sur la bande passante. À l'aide d'un contrôleur simple, des simulations sont réalisées à l'aide des outils développés pour déterminer le débattement nécessaire du mini actif qui permet la bande passante désirée. Enfin, une interface haptique capable de reproduire une poignée de main naturelle et intuitive avec un robot est présentée. Ce chapitre peut être divisé en deux aspects, i.e., la main et le bras. Ici, la main est le robot mini et le bras, le robot macro. D'abord, un prototype de main robotique est conçu et fabriqué. Inspirée de l'anatomie humaine, cette main robotique possède une paume comprimable capable d'émuler celle de l'humain ainsi que trois doigts sous-actionnés. Un pouce passif, relié au niveau de compression de la paume, complète le tout. Le contrôle de la main se fait via une position avec rétroaction, et ce, pour chacun des deux actionneurs (un pour la paume, l'autre pour les trois doigts). Ensuite, la main robotique est montée sur un manipulateur sériel collaboratif (le Kuka LWR), le bras. Ce dernier est contrôlé en impédance autour d'une trajectoire harmonique dans un plan vertical. En fonction des paramètres de la trajectoire (amplitude, fréquence, coefficients d'amortissement et de raideur), ce prototype permet de conférer une personnalité active au robot. L'expérimentation faite auprès de sujets humains permet de déterminer les valeurs considérées plus naturelles pour les différents paramètres de la trajectoire ainsi que diverses pistes à explorer pour des travaux futurs. / This thesis presents the design of a haptic interface capable of rendering a physical human-robot interaction natural and intuitive. It is a very important subject to study with the rise in collaborative robots and the ever-increasing presence of robots in everyday life. The work presented here focuses on the macro-mini architecture as haptic interface, more precisely on three important aspects to consider during the design of a macro-mini system. The first chapter explores how to physically communicate with a robot (make it understand the human's intentions) in a context of collaborative motion. In more details, the goal is to compare different methods to translate human motions (or intentions) to a robot. In this case, low impedance passive articulated shells are mounted on the links of a serial manipulator. The human operator interacts with the robot by displacing the shells. The robot's response is then to move so that its links follow the motion of their associated shell. The better the robot can follow the shells, the closest to their neutral configuration the shells can remain. The shell displacement relative to its link is considered as a desired velocity of the link. The translation of the shell displacement into joint motion then becomes an inverse kinematic problem. Different strategies can be used to solve this problem. This project then aims at comparing the efficiency of those strategies. To this end, a general study of the different strategies is performed. Then, a mathematical formalism is described, adapting said strategies to the present context. Indeed, depending on the type of shell and the position of the link in the chain, all degrees of freedom are not necessarily possible. This formalism takes these limitations into account. Then, simulations are conducted to observe the behaviour of the different strategies studied and a performance index is chosen to compare them. Afterwards, once the robot is capable of efficiently understanding the human intentions, the next step is to determine how to detect the intentions with the help of an interface and, most of all, what size should this interface have so as to communicate well. In other words, the second chapter presents an analysis of the impact that the range of motion of an active mini mechanism has on the mechanical bandwidth for the possible motions during the handling of large payloads. Indeed, the macro-mini architecture generally uses a passive mini robot, which means that the human operator feels the whole inertia of the payload. When the payload becomes sufficiently heavy, it becomes necessary for the mini robot to help the operator by working as well so as to keep the interaction natural. This means that the mini robot should then be actuated, i.e., active. It is however important that the mini robot remains backdrivable for the macro-mini architecture to work properly. Mathematical models are then presented. The limitations related to the application are described, as well as their effect on the bandwidth. With the help of a simple controller, simulations are performed with the tools developed to determine the range of motion necessary for the active mini robot which would allow the desired bandwidth. Finally, a haptic interface capable of emulating a natural and intuitive handshake with a robot is presented. This chapter can be divided into two aspects, i.e., the hand and the arm. Here, the hand is the mini robot and the arm, the macro robot. First, a robotic hand prototype is designed and constructed. Inspired by the human anatomy, this robotic hand has a compliant palm able to emulate a human palm as well as three underactuated fingers. A passive thumb, tied to the palm compression level, completes the hand. The hand control is done with position control with feedback for both actuators (one for the palm, the other for the three fingers). Then, the robotic hand is mounted on a collaborative serial manipulator (the Kuka LWR), the arm. The arm is controlled in impedance around a harmonic trajectory in a vertical plane. Depending on the parameters for the trajectory (amplitude, frequency, stiffness and damping coefficients), this prototype provides an active personality to the robot. Experimentation is conducted with human subjects to determine the values considered more natural for the different trajectory parameters as well as several improvements for the prototype in future works.

Industrial robots have been identified as one of the most effective solutions for optimising output and quality within many industries. However, there are a number of manufacturing applications involving complex tasks and inconstant components which prohibit the use of fully automated solutions in the foreseeable future. A breakthrough in robotic technologies and changes in safety legislations have supported the creation of robots that coexist and assist humans in industrial applications. It has been broadly recognised that human-robot collaborative systems would be a realistic solution as an advanced production system with wide range of applications and high economic impact. This type of system can utilise the best of both worlds, where the robot can perform simple tasks that require high repeatability while the human performs tasks that require judgement and dexterity of the human hands. Robots in such system will operate as “intelligent assistants”. In a collaborative working environment, robot and human share the same working area, and interact with each other. This level of interface will require effective ways of communication and collaboration to avoid unwanted conflicts. This project aims to create a user interface for industrial collaborative robot system through integration of current robotic technologies. The robotic system is designed for seamless collaboration with a human in close proximity. The system is capable to communicate with the human via the exchange of gestures, as well as visual signal which operators can observe and comprehend at a glance. The main objective of this PhD is to develop a Human-Robot Interface (HRI) for communication with an industrial collaborative robot during collaboration in proximity. The system is developed in conjunction with a small scale collaborative robot system which has been integrated using off-the-shelf components. The system should be capable of receiving input from the human user via an intuitive method as well as indicating its status to the user ii effectively. The HRI will be developed using a combination of hardware integrations and software developments. The software and the control framework were developed in a way that is applicable to other industrial robots in the future. The developed gesture command system is demonstrated on a heavy duty industrial robot.

Like computer architects, robot designers must address multiple, possibly competing, requirements by balancing trade-offs in terms of processing, memory, communication, and energy to satisfy design objectives. However, robot architects currently lack the design guidelines, organizing principles, rules of thumb, and tools that computer architects rely upon. This thesis takes a step in this direction, by analyzing the roles of heterogeneity and distribution in robot systems architecture. This thesis takes a systems architecture approach to the design of robot systems, and in particular, investigates the use of distributed, heterogeneous platforms to exploit locality in robot systems design. We show how multiple, distributed heterogeneous platforms can serve as general purpose robot systems for three distinct domains with different design objectives: increasing availability in a search and rescue mission, increasing flexibility and ease-of-use for a personal educational robot, and decreasing the computation and sensing resources necessary for navigation and foraging tasks.

This thesis was motivated by a study of how robots can be taught by humans, with an emphasis on allowing persons without programming skills to teach robots. The focus of this thesis was to investigate what criteria could or should be used by a robot to evaluate whether a human teacher is (or potentially could be) a good teacher in robot learning by demonstration. In effect, choosing the teacher that can maximize the benefit to the robot using learning by imitation/demonstration. The study approached this topic by taking a technology snapshot in time to see if a representative example of research laboratory robot technology is capable of assessing teaching quality. With this snapshot, this study evaluated how humans observe teaching quality to attempt to establish measurement metrics that can be transferred as rules or algorithms that are beneficial from a robot's point of view. To evaluate teaching quality, the study looked at the teacher-student relationship from a human-human interaction perspective. Two factors were considered important in defining a good teacher: engagement and immediacy. The study gathered more literature reviews relating to further detailed elements of engagement and immediacy. The study also tried to link physical effort as a possible metric that could be used to measure the level of engagement of the teachers. An investigatory experiment was conducted to evaluate which modality the participants prefer to employ in teaching a robot if the robot can be taught using voice, gesture demonstration, or physical manipulation. The findings from this experiment suggested that the participants appeared to have no preference in terms of human effort for completing the task. However, there was a significant difference in human enjoyment preferences of input modality and a marginal difference in the robot's perceived ability to imitate. A main experiment was conducted to study the detailed elements that might be used by a robot in identifying a 'good' teacher. The main experiment was conducted in two subexperiments. The first part recorded the teacher's activities and the second part analysed how humans evaluate the perception of engagement when assessing another human teaching a robot. The results from the main experiment suggested that in human teaching of a robot (human-robot interaction), humans (the evaluators) also look for some immediacy cues that happen in human-human interaction for evaluating the engagement.