人間とロボット

Enhancing Humans Trust in Robots through Explanations

Bidirectional Human-Robot Learning: Imitation and Skill Improvement

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

In order to further reduce the incisions of laparoscopic surgery and the possibility of infection, the organic combination of single-incision laparoscopic surgery (SILS) and robotics has made the degree of minimally invasive surgery further improved. A new 5-DOF single-incision laparoscopic surgery robot was designed based on Axiomatic Design Theory, whose structure consisting of the movement mechanism, the endoscope and the position and pose adjustment mechanism. The robot parts are connected in series and parallel, allowing a pivotal motion of the endoscope in the center of the robot for realizing the incision. In order to achieve a performance optimization and a dynamic control of the single-incision laparoscopic surgical robot, the kinematics and dynamic modeling and dynamic stiffness analysis of the robot are especially important. The forward kinematics equation, inverse kinematics equation and Jacobian matrix of the SILS robot are derived based on D-H method and geometric method, and the kinematics numerical simulation is carried out by Matlab. The dynamic equation of the robot is derived by Kane method. Subsequently, a numerical simulation of the robot dynamics equation is performed, with its virtual prototype utilized to set the motion plan of the robot mechanism. After robot’s dynamic simulation, the numerical changes of the driving force and torque for each robot’s moving mechanism are obtained. The performed simulation results further verify the correctness of the established robot’s dynamic model. Finally, utilizing the above methods, the dynamic stiffness model and evaluation index of the robot are determined, and the dynamic stiffness of the robot is analyzed and evaluated. The results of the kinematics, dynamic and stiffness analysis of the SILS robot further validate that the 5-DOF SILS robot has a reasonable structure, motion and sufficient stability to meet the needs of single-incision laparoscopic surgery.

Domestic service robots are nowadays widely available on the consumer market. As such, robots have begun entering people’s homes and daily lives. However, it seems that the dissemination of domestic robots has not happened as easily and widespread as it was anticipated in the first place. Little is known about the reasons why because long-term studies of ordinary people using real robots in their homes are rare. To better understand how people interact, use and accept domestic robots, studies of human-robot interaction require ecologically valid settings and the user and their needs have to come into the focus. In this dissertation, we propose to investigate the dynamics of human-robot interaction in domestic environments. We first explore the field by means of a 6-month ethnographic study of nine households. We provided each of the households with a Roomba vacuum cleaning robot. Our motivation is to understand long-term acceptance and to identify factors that can promote and hinder the integration of a domestic service robot in different types of households. We would like to find out how people’s perception of the robot, and the way they interact with it and use it, evolve over time. Furthermore, as social factors were highlighted in previous studies on technology adoption in homes, we shed light on to what extent people view Roomba and other types of domestic robots as a social entity and to what extent they anthropomorphize it. Findings of this research can be used to guide the design of user-oriented robots that have the potential to lastingly become a valuable part within the home ecology. Then, we pursue the idea of developing our own domestic robot prototype that could be used in a household with children. We imagine a playful robot that aims to motivate young children to tidy up their toys. In a first evaluation of the robot in 14 family homes, we study the effect of a proactive and reactive robot behavior on children’s interaction with the robot and their motivation to tidy up. A follow-up experiment explores the possibility to sustain children’s engagement by manipulating the robot’s behavior in such way that it appears unexpected. We further investigate how far this influences children’s perception of the robot in terms of anthropomorphism. Our findings emphasize the importance of research in ecologically valid settings in order to obtain a better understanding of human-robot interaction, advance further the design of user-oriented robots and foster the long-term acceptance of these devices.

Due to declined birthrate and the increasing aging population, shortage of caregiving labor force has become a critical issue worldwide. Introducing robotic products could provide an effective way to help older adults’ daily lives. However, previous studies indicated that older adults’ acceptance of robots was lower than younger adults. One possible reason of this lower acceptance of robots might be robot appearance. The Uncanny Valley Phenomenon (UVP) refers to the phenomenon that people rate more positively as robots become more humanlike, but only up to a certain point; as it approaches near-perfect similarity of human appearance, likeability drops and forms an uncanny valley. Nonetheless, evidence for the UVP were mainly from younger adults. We therefore examined whether the UVP is also applicable for older and middle-aged adults in the present study. We also examined whether the acceptance of functions (companion vs. service) would change based on robot appearance, and whether perceived personalities have any relation with the acceptance of robot function. We asked younger (N= 80, age 18-39), middle-aged (N= 87, age 40-59), and older (N= 88, age 60-87) adults to view each picture of a set of robot pictures selected from a totally 84 robots and evaluate their impression on each robot and the intention of use regarding robot functions. UVP was found in younger and middle-aged adults; however, older adults did not show UVP. They preferred humanlike over non-humanlike robots, regardless of robot function. Scores on each personality－except for authoritativeness－showed positive correlations with the acceptance of functions. These findings suggest that the design of assistive robots should take UVP into consideration by customizing robot appearance in accordance with the function provided to different age groups.

Robotic technologies will continue to enter new applications in addition to automated manufacturing and logistics. Once mobile robots can also operate outside of today's special facilities, they have the potential to relieve us of dirty and dangerous labor in various areas. However, for this purpose, these machines will need to be able to navigate autonomously in complex natural, urban, and industrial settings. This thesis addresses the development of locomotion skills for legged robots in challenging environments. Our work focuses on perceptive locomotion where exteroceptive sensing of the surrounding is exploited to plan and control the robot’s motion. This enables quadrupedal robots to negotiate rough terrain through carefully selected contacts. In this work, we evaluate different sensing technologies and analyze their performance for local dense terrain mapping on a mobile robot. We include special conditions such as close range objects and the influence of ambient light as we find them in real-world applications. By modeling the error characteristics of the sensors, the robot can judge the quality of the resulting terrain reconstruction. As the robot moves, the surrounding is continuously mapped to capture new areas and update regions which have changed. We contribute with a mapping framework that models the terrain from a robot-centric perspective. To this end, we present a novel approach for the error propagation from the robot's state estimation to the representation of the map. This formulation allows for robust and high-rate local mapping that is independent of a global localization method. We introduce our approach to locomotion planning, which finds safe footholds along with collision-free swing-leg motions, leveraging the generated terrain map. A nonlinear optimization finds postures that respect kinematic and stability constraints. We experimentally verify this work with torque-controllable quadrupedal robots that autonomously traverse obstacles, such as rubble, steps, gaps, and stairs without prior knowledge of the scene or any additional equipment. The locomotion planner re-plans its motion at every step in real-time, to cope with disturbances and dynamic environments. For the control of the legged robot, we contribute architecturally to the versatile and task-oriented motion execution. This method enables the robust tracking of motion plans, even with significant mismatches between the models and reality. In addition to rough terrain locomotion, we demonstrate the integration of our method for applications, such as whole-body stair climbing, manipulation, jumping, docking, inspection, payload delivery, dancing, and more. Our approach is thoroughly validated with the quadrupedal robot ANYmal in realistic long-term missions for autonomous industrial inspection and search and rescue. Finally, we extend our work with the design and implementation of a collaborative navigation framework for ground and flying robots. The ground vehicle utilizes the data captured by the flying robot to navigate uncharted…

Multi-robot systems have demonstrated strong capability in handling environmental operations. In this study, We examine how a team of robots can be utilized in covering and removing spill patches in a dynamic environment by executing three consecutive stages: deployment, coverage, and rendezvous. For the deployment problem, we aim for robot allocation based on the discreteness of the patches that need to be covered. With the deep neural network (DNN) based spill detector and remote sensing facilities such as drones with vision sensors and satellites, we are able to obtain the spill distribution in the workspace. Then, we formulate the allocation problem in a general optimization form and provide solutions using an integer linear programming (ILP) solver under several realistic constraints. After the allocation process is completed and the robot team is divided according to the number of spills, we deploy robots to their computed optimal goal positions. In the robot deployment part, control laws based on artificial potential field (APF) method are proposed and practiced on robots with a common unicycle model. For the coverage control problem, we show two strategies that are tailored for a wirelessly networked robot team. We propose strategies for coverage with and without path planning, depending on the availability of global information. Specifically, in terms of coverage with path planning, we partition the workspace from the aerial image into pieces and let each robot take care of one of the pieces. However, path-planning-based coverage relies on GPS signals or other external positioning systems, which are not applicable for indoor or GPS-denied circumstances. Therefore, we propose an asymptotic boundary shrink control that enables a collective coverage operation with the robot team. Such a strategy does not require a planned path, and because of its distributedness, it shows many advantages, including system scalability, dynamic spill adaptability, and collision avoidance. In case of a large-scale patch that poses challenges to robot connectivity maintenance during the operation, we propose a pivot-robot coverage strategy by mean of an a priori geometric tessellation (GT). In the pivot-robot-based coverage strategy, a team of robots is sent to perform complete coverage to every packing area of GT in sequence. Ultimately, the entire spill in the workspace can be covered and removed. For the rendezvous problem, we investigate the use of graph theory and propose control strategies based on network topology to motivate robots to meet at a designated or the optimal location. The rendezvous control strategies show a strong robustness to some common failures, such as mobility failure and communication failure. To expedite the rendezvous process and enable herding control in a distributed way, we propose a multi-robot multi-point rendezvous control strategy. To verify the validity of the proposed strategies, we carry out simulations in the Robotarium MATLAB platform, which is an open source swarm robotics experiment testbed, and conduct real experiments involving multiple mobile robots.

Early studies on robot machining were reported in the 1990s. Even though there are continuous worldwide researches on robot machining ever since, the potential of robot applications in machining has yet to be realized. In this paper, the authors will first look into recent development of robot machining. Such development can be roughly categorized into researches on robot machining system development, robot machining path planning, vibration/chatter analysis including path tracking and compensation, dynamic, or stiffness modeling. These researches will obviously improve the accuracy and efficiency of robot machining and provide useful references for developing robot machining systems for tasks once thought to only be capable by CNC machines. In order to advance the technology of robot machining to the next level so that more practical and competitive systems could be developed, the authors suggest that future researches on robot machining should also focus on robot machining efficiency analysis, stiffness map-based path planning, robotic arm link optimization, planning, and scheduling for a line of machining robots.

Objective To compare the clinical outcomes between minimal-invasive triangular fixation with orthopedic robot and traditional open fixation method for unilateral unstable sacral fracture patients. Methods Data of 24 consecutive patients with unilateral unstable sacral fracture who were treated from August 2014 to February 2018 were retrospectively analyzed. All patients were associated with anterior ring injuries of pelvis and received magnetic resonance of nerve (MRN) preoperatively to exclude the compression of sacral nerve by bone. All patients received surgical treatment of sacral fractures with triangular fixation and the fixation of pelvic anterior ring injuries simultaneously and two groups were divided according to the different surgical methods of sacral fractures. There were 10 males and 2 females with an average age of 36.3±1.2 years in the orthopaedic robot group. According to Dennis classification, there were 4 type I and 8 type II fractures. Two patients were associated with nerve injuries (Gibbons II 1 case, III 1 case). In traditional posterior-midline open fixation group, there were 11 males and 1 female with an average age of 38.2±1.6 years. According to Dennis classification, there were 3 type I and 9 type II fractures. Three patients were associated with nerve injuries (Gibbons II 2 cases, III 1 case). The clinical data of two group patients were collected and compared statistically. T test was used to compare the operation time, intraoperative bleeding, intraoperative fluoroscopy times, Majeed function assessment which was to evaluate the patients' clinical prognosis and healing time of fracture. χ2 test was used to compare the healing rate of fracture, accuracy assessment of fixation insertion, and Mears radiological assessment which was applied to evaluate the reduction quality of fractures. The rank sum test was used to compare Gibbons score which was applied as the index of neurological deficiency recovery. The Fisher exact test was used to compare the infection rate. Results All patients were followed up continuously for an average time of 21.2±3.2 months. The average operation time of robot group was 100.3±14.5 minutes, meanwhile the open fixation group was 202.0±18.5 min. The average intraoperative bleeding of robot group was 180.0±17.4 ml, meanwhile the open fixation group was 850.0±15.2 ml. The average intraoperative fluoroscopy time of robot group was 23.3±4.5 s, meanwhile the open fixation group was 90.0±7.7 s. All fractures were healed and no loss of reduction or fail of fixation occurred in both groups. The healing time of fracture of robot group was 8.5±1.9 months, meanwhile the open fixation group was 12.8±2.4 months. The satisfaction rates of reduction which was based on Mears-Velyvis radiological criterion of both groups were 91.7%. The accuracy rate of fixation insertion of robot group was 100% meanwhile the open fixation group was 77.78%. Majeed function assessment score of robot group was 86.2±3.4, meanwhile the open fixation group was 84.2±2.7. There was no infection occurred in robot group, meanwhile 3 patients infected in open fixation group. The Gibbons score of one patient changed from II preoperative to I postoperative and one case changed from III preoperative to II postoperative in robot group, meanwhile two patients changed from II preoperative to I postoperative and one case changed from III preoperative to II postoperative in open fixation group. The healing rate of fracture, infection rate, Majeed function assessment, Mears-Velyvis radiological evaluation criterion and Gibbons score of two groups had no significant statistical difference (P >0.05), meanwhile the operation time (t=14.99), intraoperative bleeding (t=100.46), intraoperative fluoroscopy time (t=32.13), healing time of fracture (t=4.87) and accuracy rate of fixation insertion (χ2=9.00) of orthopedic robot group were better than traditional open group and had the significant difference (P< 0.05). Conclusion The minimal-invasive triangular fixation with orthopedic robot for unilateral unstable sacral fracture had the advantages of less operation time, less intraoperative bleeding and less times of fluoroscopy, more accurate of fixation insertion and less healing time of sacral fractures compared to traditional open fixation method and should be recommended as an effective and advanced choice. Key words: Sacrum; Fractures, bone; Fracture fixation, internal; Internal fixators

In this thesis we present the design and implementation of a novel self-reconfiguring modular (SR-MR) robotic system: Roombots. We are aiming at three main applications with Roombots; locomotion through self-reconfiguration in the regular cubic 3D-lattice on structured surfaces, locomotion in non-structured environments applying central pattern generators (CPG) as the locomotion controller, and self-assembly and reconfiguration of static objects of the day-to-day environment, such as furniture. Robot assemblies from self-reconfigurable modular robots have the ability to adapt to a given task and working environment by altering their shape through a series of reconfiguration moves, and attachments and detachments between the modules. We are interested in self-reconfiguring modular robots for their shape-changing capabilities, and their distributed characteristics. We envision the following applications for Roombots: First, self-reconfiguration in a structured 3D lattice, i.e. a floor and walls equipped with connectors. Embedded connectors can provide pivot points for locomotion of SR-MR assemblies, and docking and recharging places for our adaptive furniture pieces. Second, our proposed concept for locomotion control of modular robots on non-structured ground are central pattern generators. Third, we would like to build adaptive and versatile furniture from modular robots and light-weight elements. In the following we are able to count more than 60 modular robotic systems, developed over the last two decades. However we were unable to identify an existing system which could provide us with all desired kinematic and geometric capabilities. This led us to design and implement a novel self-reconfiguring modular robotic system: Roombots. To tackle the module design we attempt to identify both meaningful design parameters from existing modular robots, and essential features for our applications. The combination of both leads to the kinematic and geometric description of the Roombots modules, and eventually to its implementation. In order to be able to assemble furniture from our Roombots units in the future, we need a reconfiguration framework which supports the specific requirements of Roombots. Metamodules made of two units attached in-series are attracted and guided by a virtual force-field, they use broadcast signals, look-up tables of collision clouds and simple assumptions about their near environment to reach their seeding positions, which are currently hand coded. For the task of locomotion in non-structured environments we propose a framework for learning to move with modular robots using central pattern generators and online optimization. The distributed implementation of CPGs offers an ideal substrate for producing locomotion patterns and for online learning, and an optimization framework for fast learning.

We consider a distributed system consisting of autonomous mobile computing entities called robots moving in the three-dimensional space (3D-space). The robots are anonymous, oblivious, fully-synchronous and have neither any access to the global coordinate system nor any explicit communication medium. Each robot cooperates with other robots by observing the positions of other robots in its local coordinate system. One of the most fundamental agreement problems in 3D-space is the plane formation problem that requires the robots to land on a common plane, that is not predefined. This problem is not always solvable because of the impossibility of symmetry breaking. While existing results assume that the robots agree on the handedness of their local coordinate systems, we remove the assumption and consider the robots without chirality. The robots without chirality can never break the symmetry consisting of rotation symmetry and reflection symmetry. Such symmetry in 3D-space is fully described by 17 symmetry types each of which forms a group. We extend the notion of symmetricity [Suzuki and Yamashita, SIAM J. Compt. 1999] [Yamauchi et al., PODC 2016] to cover these 17 symmetry groups. Then we give a characterization of initial configurations from which the fully-synchronous robots without chirality can form a plane in terms of symmetricity.

Recently, the electronic field develop rapidly especially in creating robotic. Many competitions have conducted to compare technology, from hobbies until robotic industries are never bored to talk. Industrially, automotive robotic needed in order to make the work more efficient, then it save cost of a production. Automotive robotic used also in entering dangerous area, that there is unable to handle by human for safety reason.In this paper, there are two automotive robotic that created, which able to follow line and stop when it detects any obstruction at front side. This robotic designed by Microcontroller AT89551, ultrasonic sensor as obstruction detector and it also have four infrared and photodiode sensors as line detector which through by feet. Automotive robotic will run following designed line, if there is any obstruction then it will stop as long as those obstructions are there and it run again if the ultrasonic sensor do not detect obstruction anymore. By regulating PWM signal (Pulse Width Modulation) from microcontroller, then the automotive robotic can walk straightly, turn and will stop if there are any obstruction detected.From the experiment and test which performed, robot is able to pass the turn with 60 cm of diameter, while for movement fork it is only perform straightly and stopped, exactly if robot finding the obstruction it can stoped in the distance 40 cm. It shows that there is lack of sensor, which used in designing, then for particular fork, robotic is only able to stop and walk straightly. Keywords: Microcontroller AT89S51, Lines Follower, Obstruction Detector

The safety and reliability of robots, like other engineering products, have been considered as important issues in many countries since the growing robot technology entered the industry. An industrial robot must be safe and reliable so that it does not lead to unsafe situations and high maintenance expenses. The growing application of industrial robots in some of the industries of Iran, and the nature of their activities (vast work environment, unpredictable movements, and the nature of controlling its computer program) will create a unique challenge in occupational safety. Consecutive failures of a robot will cause an industry to suffer from great expenses. This study seeks to develop a safety analysis model for industrial robots. Due to the importance of this issue and the dearth of studies done in this regard, this study intended to develop a safety analysis model for industrial robots based on Markov chain. Then, this model was applied to the robots in Haierplast Company; and finally, the results were analyzed. The findings of this study include the computation of danger rate, probabilities, reliability, the average of failure time, and the repair rate of the safety system of robots. Keywords: robot safety, reliability, severity-frequency index of an event

Robots had a great impact on the manufacturing industry ever since the early seventies when companies such as KUKA and ABB started deploying their first industrial robots. These robots merely performed very specific tasks in specific ways within well-defined environments. Still, they proved to be very useful as they could exceed human performance at these tasks. However, in order to enable robots to enter our daily life, they need to become more versatile and need to operate in much less structured environments. This thesis is partly devoted to stretching these limitations by means of learning, namely imitation learning (IL) and inverse reinforcement learning (IRL). Reinforcement learning (RL) is a powerful approach to enable robots to solve a task in an unknown environment. The practitioner describes a desired behavior by specifying a reward function and the robot autonomously interacts with the environment in order to find a control policy that generates high accumulated reward. However, RL is not suitable for teaching new tasks by non-experts because specifying appropriate reward functions can be difficult. Demonstrating the desired behavior is often easier for non-experts. Imitation learning can be used in order to enable the robot to reproduce the demonstrations. However, without explicitly inferring and modeling the intentions of the demonstrations, it can become difficult to solve the task for unseen situations. Inverse reinforcement learning (IRL) therefore aims to infer a reward function from the demonstrations, such that optimizing this reward function yields the desired behavior even for different situations. This thesis introduces a unifying approach to solve the inverse reinforcement learning problem in the same way as the reinforcement learning problem. This is achieved by framing both problems as information projection problems, i.e., we strive to minimize the relative entropy between a probabilistic model of the robot behavior and a given desired distribution. Furthermore, a trust region on the robot behavior is used to stabilize the optimization. For inverse reinforcement learning, the desired distribution is implicitly given by the expert demonstrations. The resulting optimization can be efficiently solved using state-of-the-art reinforcement learning methods. For reinforcement learning, the log-likelihood of the desired distribution is given by the reward function. The resulting optimization problem corresponds to a standard reinforcement learning formulation, except for an additional objective of maximizing the entropy of the robot behavior. This entropy objective adds little overhead to the optimization, but can lead to better exploration and more diversified policies. Trust-region I-projections are not only useful for training robots, but can also be applied to other machine learning problems. I-projections are typically used for variational inference, in order to approximate an intractable distribution by a simpler model. However, the resulting optimization problems are usually optimized based on stochastic gradient descent which often suffers from high variance in the gradient estimates. As trust-region I-projections where shown to be effective for reinforcement learning and inverse reinforcement learning, this thesis also explores their use for variational inference. More specifically, trust-region I-projections are investigated for the problem of approximating an intractable distribution by a Gaussian mixture model (GMM) with an adaptive number of components. GMMs are highly desirable for variational inference because they can yield arbitrary accurate approximations while inference from GMMs is still relatively cheap. In order to make learning the GMM feasible, we derive a lower bound that enables us to decompose the objective function. The optimization can then be performed by iteratively updating individual components using a technique from reinforcement learning. The resulting method is capable of learning approximations of significantly higher quality than existing variational inference methods. Due to the similarity of the underlying optimization problems, the insights gained from our variational inference method are also useful for IL and IRL. Namely, a similar lower bound can be applied also for the I-projection formulation of imitation learning. However, whereas for variational inference the lower bound serves to decompose the objective function, for imitation learning it allows us to provide a reward signal to the robot that does not depend on its behavior. Compared to reward functions that are relative to the current behavior of the robot---which are typical for popular adversarial methods---behavior-independent reward functions have the advantages that we can show convergence even for greedy optimization. Furthermore, behavior-independent reward functions solve the inverse reinforcement learning problem, thereby closing the gap between imitation learning and IRL. However, algorithms derived from our non-adversarial formulation are actually very similar to existing AIL methods, and we can even show that adversarial inverse reinforcement learning (AIRL) is indeed an instance of our formulation. AIRL was derived from an adversarial formulation, and we point out several problems of that derivation. In contrast, we show that AIRL can be straightforwardly derived from out non-adversarial formulation. Furthermore, we demonstrate that the non-adversarial formulation can be also used to derive novel algorithms by presenting a non-adversarial method for offline imitation learning.