scispace - formally typeset
Search or ask a question

Showing papers on "Humanoid robot published in 2013"


Journal ArticleDOI
TL;DR: This paper studies the properties, structure and computation schemes for the centroidal momentum matrix (CMM), which projects the generalized velocities of a humanoid robot to its spatial centroidAl momentum, and introduces the new concept of “average spatial velocity” of the humanoid that encompasses both linear and angular components and results in a novel decomposition of the kinetic energy.
Abstract: The center of mass (CoM) of a humanoid robot occupies a special place in its dynamics. As the location of its effective total mass, and consequently, the point of resultant action of gravity, the CoM is also the point where the robot's aggregate linear momentum and angular momentum are naturally defined. The overarching purpose of this paper is to refocus our attention to centroidal dynamics: the dynamics of a humanoid robot projected at its CoM. In this paper we specifically study the properties, structure and computation schemes for the centroidal momentum matrix (CMM), which projects the generalized velocities of a humanoid robot to its spatial centroidal momentum. Through a transformation diagram we graphically show the relationship between this matrix and the well-known joint-space inertia matrix. We also introduce the new concept of "average spatial velocity" of the humanoid that encompasses both linear and angular components and results in a novel decomposition of the kinetic energy. Further, we develop a very efficient $$O(N)$$ O ( N ) algorithm, expressed in a compact form using spatial notation, for computing the CMM, centroidal momentum, centroidal inertia, and average spatial velocity. Finally, as a practical use of centroidal dynamics we show that a momentum-based balance controller that directly employs the CMM can significantly reduce unnecessary trunk bending during balance maintenance against external disturbance.

400 citations


Journal ArticleDOI
TL;DR: When the robot used co-verbal gestures during interaction, it was anthropomorphized more, participants perceived it as more likable, reported greater shared reality with it, and showed increased future contact intentions than when the robot gave instructions without gestures.
Abstract: Previous work has shown that non-verbal behaviors affect anthropomorphic inferences about artificial communicators such as virtual agents or social robots In an experiment with a humanoid robot we investigated the effects of the robot’s hand and arm gestures on the perception of humanlikeness, likability of the robot, shared reality, and future contact intentions after interacting with the robot For this purpose, the speech-accompanying non-verbal behaviors of the humanoid robot were manipulated in three experimental conditions: (1) no gesture, (2) congruent co-verbal gesture, and (3) incongruent co-verbal gesture We hypothesized higher ratings on all dependent measures in the two multimodal (ie, speech and gesture) conditions compared to the unimodal (ie, speech only) condition The results confirm our predictions: when the robot used co-verbal gestures during interaction, it was anthropomorphized more, participants perceived it as more likable, reported greater shared reality with it, and showed increased future contact intentions than when the robot gave instructions without gestures Surprisingly, this effect was particularly pronounced when the robot’s gestures were partly incongruent with speech, although this behavior negatively affected the participants’ task-related performance These findings show that communicative non-verbal behaviors displayed by robotic systems affect anthropomorphic perceptions and the mental models humans form of a humanoid robot during interaction

282 citations


Journal ArticleDOI
TL;DR: The task-function approach is extended to handle the full dynamics of the robot multibody along with any constraint written as equality or inequality of the state and control variables to keep a low computation cost.
Abstract: The most widely used technique for generating whole-body motions on a humanoid robot accounting for various tasks and constraints is inverse kinematics. Based on the task-function approach, this class of methods enables the coordination of robot movements to execute several tasks in parallel and account for the sensor feedback in real time, thanks to the low computation cost. To some extent, it also enables us to deal with some of the robot constraints (e.g., joint limits or visibility) and manage the quasi-static balance of the robot. In order to fully use the whole range of possible motions, this paper proposes extending the task-function approach to handle the full dynamics of the robot multibody along with any constraint written as equality or inequality of the state and control variables. The definition of multiple objectives is made possible by ordering them inside a strict hierarchy. Several models of contact with the environment can be implemented in the framework. We propose a reduced formulation of the multiple rigid planar contact that keeps a low computation cost. The efficiency of this approach is illustrated by presenting several multicontact dynamic motions in simulation and on the real HRP-2 robot.

226 citations


Proceedings ArticleDOI
06 May 2013
TL;DR: A novel systematic method to optimally tune the joint elasticity of multi-dof SEA robots based on resonance analysis and energy storage maximization criteria forms one of the key contributions of this work.
Abstract: The incorporation of passive compliance in robotic systems could improve their performance during interactions and impacts, for energy storage and efficiency, and for general safety for both the robots and humans. This paper presents the recently developed COMpliant huMANoid COMAN. COMAN is actuated by passive compliance actuators based on the series elastic actuation principle (SEA). The design and implementation of the overall body of the robot is discussed including the realization of the different body segments and the tuning of the joint distributed passive elasticity. This joint stiffness tuning is a critical parameter in the performance of compliant systems. A novel systematic method to optimally tune the joint elasticity of multi-dof SEA robots based on resonance analysis and energy storage maximization criteria forms one of the key contributions of this work. The paper will show this method being applied to the selection of the passive elasticity of COMAN legs. The first completed robot prototype is presented accompanied by experimental walking trials to demonstrate its operation.

219 citations


Journal ArticleDOI
TL;DR: The Intention-Driven Dynamics Model is proposed to probabilistically model the generative process of movements that are directed by the intention and allows the intention to be inferred from observed movements using Bayes’ theorem.
Abstract: Intention inference can be an essential step toward efficient human-robot interaction. For this purpose, we propose the Intention-Driven Dynamics Model (IDDM) to probabilistically model the generative process of movements that are directed by the intention. The IDDM allows the intention to be inferred from observed movements using Bayes' theorem. The IDDM simultaneously finds a latent state representation of noisy and high-dimensional observations, and models the intention-driven dynamics in the latent states. As most robotics applications are subject to real-time constraints, we develop an efficient online algorithm that allows for real-time intention inference. Two human-robot interaction scenarios, i.e. target prediction for robot table tennis and action recognition for interactive humanoid robots, are used to evaluate the performance of our inference algorithm. In both intention inference tasks, the proposed algorithm achieves substantial improvements over support vector machines and Gaussian processes.

157 citations


Journal ArticleDOI
TL;DR: This work proposes a method to plan optimal whole-body dynamic motion in multi-contact non-gaited transitions using a B-spline time parameterization for the active joints and addresses the problem of the balance within the optimization problem.
Abstract: We propose a method to plan optimal whole-body dynamic motion in multi-contact non-gaited transitions. Using a B-spline time parameterization for the active joints, we turn the motion-planning problem into a semi-infinite programming formulation that is solved by nonlinear optimization techniques. Our main contribution lies in producing constraint-satisfaction guaranteed motions for any time grid. Indeed, we use Taylor series expansion to approximate the dynamic and kinematic models over fixed successive time intervals, and transform the problem (constraints and cost functions) into time polynomials which coefficients are function of the optimization variables. The evaluation of the constraints turns then into computation of extrema (over each time interval) that are given to the solver. We also account for collisions and self-collisions constraints that have not a closed-form expression over the time. We address the problem of the balance within the optimization problem and demonstrate that generating whole-body multi-contact dynamic motion for complex tasks is possible and can be tractable, although still time consuming. We discuss thoroughly the planning of a sitting motion with the HRP-2 humanoid robot and assess our method with several other complex scenarios.

146 citations


Proceedings ArticleDOI
01 Oct 2013
TL;DR: An integrated system based on real-time model-predictive control (MPC) applied to the full dynamics of the robot, which is possible due to the speed of the new physics engine (MuJoCo), the efficiency of the trajectory optimization algorithm, and the contact smoothing methods developed for the purpose of control optimization.
Abstract: Generating diverse behaviors with a humanoid robot requires a mix of human supervision and automatic control. Ideally, the user's input is restricted to high-level instruction and guidance, and the controller is intelligent enough to accomplish the tasks autonomously. Here we describe an integrated system that achieves this goal. The automatic controller is based on real-time model-predictive control (MPC) applied to the full dynamics of the robot. This is possible due to the speed of our new physics engine (MuJoCo), the efficiency of our trajectory optimization algorithm, and the contact smoothing methods we have developed for the purpose of control optimization. In our system, the operator specifies subtasks by selecting from a menu of predefined cost functions, and optionally adjusting the mixing weights of the different cost terms in runtime. The resulting composite cost is sent to the MPC machinery which constructs a new locally-optimal time-varying linear feedback control law once every 30 msec, while planning 500 msec into the future. This control law is evaluated at 1 kHz to generate control signals for the robot, until the next control law becomes available. Performance is illustrated on a subset of the tasks from the DARPA Virtual Robotics Challenge.

144 citations


Journal ArticleDOI
TL;DR: This study proposes a framework for optimization of torque and impedance profiles in order to maximize task performance, which is tuned to the complex hardware and incorporating real-world actuation constraints.
Abstract: Anthropomorphic robots that aim to approach human performance agility and efficiency are typically highly redundant not only in their kinematics but also in actuation. Variable-impedance actuators, used to drive many of these devices, are capable of modulating torque and impedance (stiffness and/or damping) simultaneously, continuously, and independently. These actuators are, however, nonlinear and assert numerous constraints, e.g., range, rate, and effort limits on the dynamics. Finding a control strategy that makes use of the intrinsic dynamics and capacity of compliant actuators for such redundant, nonlinear, and constrained systems is nontrivial. In this study, we propose a framework for optimization of torque and impedance profiles in order to maximize task performance, which is tuned to the complex hardware and incorporating real-world actuation constraints. Simulation study and hardware experiments 1) demonstrate the effects of actuation constraints during impedance control, 2) show applicability of the present framework to simultaneous torque and temporal stiffness optimization under constraints that are imposed by real-world actuators, and 3) validate the benefits of the proposed approach under experimental conditions.

141 citations


Proceedings ArticleDOI
01 Nov 2013
TL;DR: This paper presents new methods to control highspeed running in a simulated humanoid robot at speeds of up to 6.5 m/s using a 3D spring-loaded inverted pendulum (SLIP) template model, the first time that a SLIP model has been embedded into a whole-body humanoid model.
Abstract: This paper presents new methods to control highspeed running in a simulated humanoid robot at speeds of up to 6.5 m/s. We present methods to generate compliant target CoM dynamics through the use of a 3D spring-loaded inverted pendulum (SLIP) template model. A nonlinear least-squares optimizer is used to find periodic trajectories of the 3D-SLIP offline, while a local deadbeat SLIP controller provides reference CoM dynamics online at real-time rates to correct for tracking errors and disturbances. The local deadbeat controller employs common foot placement strategies that are automatically generated by a local analysis of the 3D-SLIP apex return map. A task-space controller is then applied online to select whole-body joint torques which embed these target dynamics into the humanoid. Despite the body of work on the 2D and 3D-SLIP models, to the best of the authors' knowledge, this is the first time that a SLIP model has been embedded into a whole-body humanoid model. When running at 3.5 m/s, the controller is shown to reject lateral disturbances of 40 N·s applied at the waist. A final demonstration shows the capability of the controller to stabilize running at 6.5 m/s, which is comparable with the speed of an Olympian in the 5000 meter run.

140 citations


Journal ArticleDOI
01 Mar 2013
TL;DR: This work develops an adaptive and individualized robot-mediated technology that intelligently adapts itself in an individualized manner to generate prompts and reinforcements with potential to promote skills in the ASD core deficit area of early social orienting.
Abstract: Emerging technology, especially robotic technology, has been shown to be appealing to children with autism spectrum disorders (ASD). Such interest may be leveraged to provide repeatable, accurate and individualized intervention services to young children with ASD based on quantitative metrics. However, existing robot-mediated systems tend to have limited adaptive capability that may impact individualization. Our current work seeks to bridge this gap by developing an adaptive and individualized robot-mediated technology for children with ASD. The system is composed of a humanoid robot with its vision augmented by a network of cameras for real-time head tracking using a distributed architecture. Based on the cues from the child's head movement, the robot intelligently adapts itself in an individualized manner to generate prompts and reinforcements with potential to promote skills in the ASD core deficit area of early social orienting. The system was validated for feasibility, accuracy, and performance. Results from a pilot usability study involving six children with ASD and a control group of six typically developing (TD) children are presented.

127 citations


Proceedings ArticleDOI
06 May 2013
TL;DR: This work presents an approach of inverting such precomputed reachability representations in order to generate suitable robot base positions for grasping and generates a distribution in SE(2), the cross-space consisting of 2D position and 1D orientation, that describes potential robot base poses together with a quality index.
Abstract: Having a representation of the capabilities of a robot is helpful when online queries, such as solving the inverse kinematics (IK) problem for grasping tasks, must be processed efficiently in the real world. When workspace representations, e.g. the reachability of an arm, are considered, additional quality information such as manipulability or self-distance can be employed to enrich the spatial data. In this work we present an approach of inverting such precomputed reachability representations in order to generate suitable robot base positions for grasping. Compared to existing works, our approach is able to generate a distribution in SE(2), the cross-space consisting of 2D position and 1D orientation, that describes potential robot base poses together with a quality index. We show how this distribution can be queried quickly in order to find oriented base poses from which a target grasping pose is reachable without collisions. The approach is evaluated in simulation using the humanoid robot ARMAR-III [1] and an extension is presented that allows to find suitable base poses for trajectory execution.

Proceedings ArticleDOI
06 May 2013
TL;DR: By equipping the PR2 humanoid robot with state-of-the-art biomimetic tactile sensors that measure temperature, pressure, and fingertip deformations, this research created a platform uniquely capable of feeling the physical properties of everyday objects.
Abstract: Delivering on the promise of real-world robotics will require robots that can communicate with humans through natural language by learning new words and concepts through their daily experiences. Our research strives to create a robot that can learn the meaning of haptic adjectives by directly touching objects. By equipping the PR2 humanoid robot with state-of-the-art biomimetic tactile sensors that measure temperature, pressure, and fingertip deformations, we created a platform uniquely capable of feeling the physical properties of everyday objects. The robot used five exploratory procedures to touch 51 objects that were annotated by human participants with 34 binary adjective labels. We present both static and dynamic learning methods to discover the meaning of these adjectives from the labeled objects, achieving average F1 scores of 0.57 and 0.79 on a set of eight previously unfelt items.

Journal ArticleDOI
TL;DR: A feedback controller that allows MABEL, which is a kneed planar bipedal robot with 1-m-long legs, to accommodate terrain that presents large unexpected increases and decreases in height is presented.
Abstract: This paper presents a feedback controller that allows MABEL, which is a kneed planar bipedal robot with 1-m-long legs, to accommodate terrain that presents large unexpected increases and decreases in height. The robot is provided no information regarding where the change in terrain height occurs and by how much. A finite-state machine is designed that manages transitions among controllers for flat-ground walking, stepping-up and -down, and a trip reflex. If the robot completes a step, the depth of a step-down or the height of a step-up can be immediately estimated at impact from the lengths of the legs and the angles of the robot’s joints. The change in height can be used to invoke a proper control response. On the other hand, if the swing leg impacts an obstacle during a step, or has a premature impact with the ground, a trip reflex is triggered on the basis of specially designed contact switches on the robot’s shins, contact switches at the end of each leg, and the current configuration of the robot. The design of each control mode and the transition conditions among them are presented. This paper concludes with experimental results of MABEL (blindly) accommodating various types of platforms, including ascent of a 12.5-cm-high platform, stepping-off an 18.5-cm-high platform, and walking over a platform with multiple ascending and descending steps.

Journal ArticleDOI
Bongjae Choi1, Sungho Jo1
04 Sep 2013-PLOS ONE
TL;DR: An approach that combines the P300 potential, the steady state visually evoked potential (SSVEP), and event related de-synchronization (ERD) to solve a complicated multi-task problem consisting of humanoid robot navigation and control along with object recognition using a low-cost BCI system is described.
Abstract: This paper describes a hybrid brain-computer interface (BCI) technique that combines the P300 potential, the steady state visually evoked potential (SSVEP), and event related de-synchronization (ERD) to solve a complicated multi-task problem consisting of humanoid robot navigation and control along with object recognition using a low-cost BCI system. Our approach enables subjects to control the navigation and exploration of a humanoid robot and recognize a desired object among candidates. This study aims to demonstrate the possibility of a hybrid BCI based on a low-cost system for a realistic and complex task. It also shows that the use of a simple image processing technique, combined with BCI, can further aid in making these complex tasks simpler. An experimental scenario is proposed in which a subject remotely controls a humanoid robot in a properly sized maze. The subject sees what the surrogate robot sees through visual feedback and can navigate the surrogate robot. While navigating, the robot encounters objects located in the maze. It then recognizes if the encountered object is of interest to the subject. The subject communicates with the robot through SSVEP and ERD-based BCIs to navigate and explore with the robot, and P300-based BCI to allow the surrogate robot recognize their favorites. Using several evaluation metrics, the performances of five subjects navigating the robot were quite comparable to manual keyboard control. During object recognition mode, favorite objects were successfully selected from two to four choices. Subjects conducted humanoid navigation and recognition tasks as if they embodied the robot. Analysis of the data supports the potential usefulness of the proposed hybrid BCI system for extended applications. This work presents an important implication for the future work that a hybridization of simple BCI protocols provide extended controllability to carry out complicated tasks even with a low-cost system.

Proceedings ArticleDOI
03 Mar 2013
TL;DR: The model validated that individuals preferred more to interact with a robot that had the same personality with theirs and that an adapted mixed robot's behavior was more engaging and effective than a speech only robot's behaviors in an interaction.
Abstract: Robots are more and more present in our daily life; they have to move into human-centered environments, to interact with humans, and to obey some social rules so as to produce an appropriate social behavior in accordance with human's profile (i.e., personality, state of mood, and preferences). Recent researches discussed the effect of personality traits on the verbal and nonverbal production, which plays a major role in transferring and understanding messages in a social interaction between a human and a robot. The characteristics of the generated gestures (e.g., amplitude, direction, rate, and speed) during the nonverbal communication can differ according to the personality trait, which, similarly, influences the verbal content of the human speech in terms of verbosity, repetitions, etc. Therefore, our research tries to map a human's verbal behavior to a corresponding combined robot's verbal-nonverbal behavior based on the personality dimensions of the interacting human. The system estimates first the interacting human's personality traits through a psycholinguistic analysis of the spoken language, then it uses PERSONAGE natural language generator that tries to generate a corresponding verbal language to the estimated personality traits. Gestures are generated by using BEAT toolkit, which performs a linguistic and contextual analysis of the generated language relying on rules derived from extensive research into human conversational behavior. We explored the human-robot personality matching aspect and the differences of the adapted mixed robot's behavior (gesture and speech) over the adapted speech only robot's behavior in an interaction. Our model validated that individuals preferred more to interact with a robot that had the same personality with theirs and that an adapted mixed robot's behavior (gesture and speech) was more engaging and effective than a speech only robot's behavior. Our experiments were done with Nao robot.

Journal ArticleDOI
TL;DR: A planner for underactuated hyper-redundant robots, such as humanoid robots, for which the movement can only be initiated by taking contacts with the environment is presented.

Journal ArticleDOI
TL;DR: The overall development is integrated with a modern humanoid robot platform under its Linux C++ SDKs and shows great potential in developing personalised intelligent agents/robots with emotion and social intelligence.
Abstract: Automatic perception of human affective behaviour from facial expressions and recognition of intentions and social goals from dialogue contexts would greatly enhance natural human robot interaction. This research concentrates on intelligent neural network based facial emotion recognition and Latent Semantic Analysis based topic detection for a humanoid robot. The work has first of all incorporated Facial Action Coding System describing physical cues and anatomical knowledge of facial behaviour for the detection of neutral and six basic emotions from real-time posed facial expressions. Feedforward neural networks (NN) are used to respectively implement both upper and lower facial Action Units (AU) analysers to recognise six upper and 11 lower facial actions including Inner and Outer Brow Raiser, Lid Tightener, Lip Corner Puller, Upper Lip Raiser, Nose Wrinkler, Mouth Stretch etc. An artificial neural network based facial emotion recogniser is subsequently used to accept the derived 17 Action Units as inputs to decode neutral and six basic emotions from facial expressions. Moreover, in order to advise the robot to make appropriate responses based on the detected affective facial behaviours, Latent Semantic Analysis is used to focus on underlying semantic structures of the data and go beyond linguistic restrictions to identify topics embedded in the users’ conversations. The overall development is integrated with a modern humanoid robot platform under its Linux C++ SDKs. The work presented here shows great potential in developing personalised intelligent agents/robots with emotion and social intelligence.

Journal ArticleDOI
TL;DR: The replication of the human hand's functionality and appearance is one of the main reasons for the development of robot hands and the design solutions, which are well suited to a single domain, might not be readily taken as general guidelines.
Abstract: The replication of the human hand's functionality and appearance is one of the main reasons for the development of robot hands. Despite 40 years of research in the field [1], the reproduction of human capabilities, in terms of dexterous manipulation, still seems unachievable by the state-of-the-art technologies. From a design perspective, even defining the optimal functionalities of a robotic end-effector is quite a challenging task since possible applications of these devices span industrial robotics, humanoid robotics, rehabilitation medicines, and prosthetics, to name a few. Therefore, it is reasonable to think that the design solutions, which are well suited to a single domain, might not be readily taken as general guidelines. For example, industrial manipulators are often equipped with basic grippers, which are conceived so as to increase the throughput and the reliability, and are assumed to operate in structured environments. In this case, the enhanced manipulation skills and the subsequent cost increases must be carefully motivated by the application requirements.

Proceedings ArticleDOI
01 Oct 2013
TL;DR: This work presents an optimization based real-time walking controller for a full size humanoid robot that is capable of walking on rough terrain, and also achieves longer foot steps, faster walking speed, heel-strike and toe push-off.
Abstract: We present an optimization based real-time walking controller for a full size humanoid robot. The controller consists of two levels of optimization, a high level trajectory optimizer that reasons about center of mass and swing foot trajectories, and a low level controller that tracks those trajectories by solving a floating base full body inverse dynamics problem using Quadratic Programming. Our controller is capable of walking on rough terrain, and also achieves longer foot steps, faster walking speed, heel-strike and toe push-off. Results are demonstrated with Boston Dynamics' Atlas robot in simulation.

Journal ArticleDOI
TL;DR: First, a randomized algorithm for constrained motion planning is presented, that is used to generate collision-free statically balanced paths solving manipulation tasks, and it is shown that dynamic walking makes humanoid robots small-space controllable.
Abstract: This paper presents a general method for planning collision-free whole-body walking motions for humanoid robots. First, we present a randomized algorithm for constrained motion planning, that is used to generate collision-free statically balanced paths solving manipulation tasks. Then, we show that dynamic walking makes humanoid robots small-space controllable. Such a property allows to easily transform collision-free statically balanced paths into collision-free dynamically balanced trajectories. It leads to a sound algorithm which has been applied and evaluated on several problems where whole-body planning and walk are needed, and the results have been validated on a real HRP-2 robot.

Journal ArticleDOI
27 May 2013-PLOS ONE
TL;DR: This paper studies the use of the predictive information (PI) of the sensorimotor process as a driving force to generate behavior and introduces the time-local predicting information (TiPI) which allows for exact results to be derived together with explicit update rules for the parameters of the controller in the dynamical systems framework.
Abstract: Information theory is a powerful tool to express principles to drive autonomous systems because it is domain invariant and allows for an intuitive interpretation. This paper studies the use of the predictive information (PI), also called excess entropy or effective measure complexity, of the sensorimotor process as a driving force to generate behavior. We study nonlinear and nonstationary systems and introduce the time-local predicting information (TiPI) which allows us to derive exact results together with explicit update rules for the parameters of the controller in the dynamical systems framework. In this way the information principle, formulated at the level of behavior, is translated to the dynamics of the synapses. We underpin our results with a number of case studies with high-dimensional robotic systems. We show the spontaneous cooperativity in a complex physical system with decentralized control. Moreover, a jointly controlled humanoid robot develops a high behavioral variety depending on its physics and the environment it is dynamically embedded into. The behavior can be decomposed into a succession of low-dimensional modes that increasingly explore the behavior space. This is a promising way to avoid the curse of dimensionality which hinders learning systems to scale well.

Proceedings ArticleDOI
06 May 2013
TL;DR: GSL is introduced, an iterative optimization framework for speeding up robot learning using an imperfect simulator, and is fully implemented and validated on the task of learning to walk using an Aldebaran Nao humanoid robot.
Abstract: Simulation is often used in research and industry as a low cost, high efficiency alternative to real model testing. Simulation has also been used to develop and test powerful learning algorithms. However, parameters learned in simulation often do not translate directly to the application, especially because heavy optimization in simulation has been observed to exploit the inevitable simulator simplifications, thus creating a gap between simulation and application that reduces the utility of learning in simulation.This paper introduces Grounded Simulation Learning (GSL), an iterative optimization framework for speeding up robot learning using an imperfect simulator. In GSL, a behavior is developed on a robot and then repeatedly: 1) the behavior is optimized in simulation; 2) the resulting behavior is tested on the real robot and compared to the expected results from simulation, and 3) the simulator is modified, using a machine-learning approach to come closer in line with reality. This approach is fully implemented and validated on the task of learning to walk using an Aldebaran Nao humanoid robot. Starting from a set of stable, hand-coded walk parameters, four iterations of this three-step optimization loop led to more than a 25% increase in the robot's walking speed.

Proceedings ArticleDOI
01 Oct 2013
TL;DR: Recent approaches on learning parameterizable skills based on dynamical movement primitives based on DMPs are generalized, such that task parameters are also passed as inputs to the function approximator of the DMP.
Abstract: One of the long-term challenges of programming by demonstration is achieving generality, i.e. automatically adapting the reproduced behavior to novel situations. A common approach for achieving generality is to learn parameterizable skills from multiple demonstrations for different situations. In this paper, we generalize recent approaches on learning parameterizable skills based on dynamical movement primitives (DMPs), such that task parameters are also passed as inputs to the function approximator of the DMP. This leads to a more general, flexible, and compact representation of parameterizable skills, as demonstrated by our empirical evaluation on the iCub and Meka humanoid robots.

Book ChapterDOI
TL;DR: In this article, a method based on 3D Histograms of Scene Flow (3DHOFs) and Global Histogram of Oriented Gradient (GHOGs) is proposed for one-shot action modeling and recognition.
Abstract: Sparsity has been showed to be one of the most important properties for visual recognition purposes. In this paper we show that sparse representation plays a fundamental role in achieving one-shot learning and real-time recognition of actions. We start off from RGBD images, combine motion and appearance cues and extract state-of-the-art features in a computationally efficient way. The proposed method relies on descriptors based on 3D Histograms of Scene Flow (3DHOFs) and Global Histograms of Oriented Gradient (GHOGs); adaptive sparse coding is applied to capture high-level patterns from data. We then propose a simultaneous on-line video segmentation and recognition of actions using linear SVMs. The main contribution of the paper is an effective real-time system for one-shot action modeling and recognition; the paper highlights the effectiveness of sparse coding techniques to represent 3D actions. We obtain very good results on three different data sets: a benchmark data set for one-shot action learning (the ChaLearn Gesture Data Set), an in-house data set acquired by a Kinect sensor including complex actions and gestures differing by small details, and a data set created for human-robot interaction purposes. Finally we demonstrate that our system is effective also in a human-robot interaction setting and propose a memory game, "All Gestures You Can", to be played against a humanoid robot.

Journal ArticleDOI
22 Mar 2013-PLOS ONE
TL;DR: The quantitative behaviour analysis reveal that the most notable difference between the interviews with KASPAR and the human were the duration of the interviews, the eye gaze directed towards the different interviewers, and the response time of the interviewers.
Abstract: Robots have been used in a variety of education, therapy or entertainment contexts. This paper introduces the novel application of using humanoid robots for robot-mediated interviews. An experimental study examines how children’s responses towards the humanoid robot KASPAR in an interview context differ in comparison to their interaction with a human in a similar setting. Twenty-one children aged between 7 and 9 took part in this study. Each child participated in two interviews, one with an adult and one with a humanoid robot. Measures include the behavioural coding of the children’s behaviour during the interviews and questionnaire data. The questions in these interviews focused on a special event that had recently taken place in the school. The results reveal that the children interacted with KASPAR very similar to how they interacted with a human interviewer. The quantitative behaviour analysis reveal that the most notable difference between the interviews with KASPAR and the human were the duration of the interviews, the eye gaze directed towards the different interviewers, and the response time of the interviewers. These results are discussed in light of future work towards developing KASPAR as an ‘interviewer’ for young children in application areas where a robot may have advantages over a human interviewer, e.g. in police, social services, or healthcare applications.

Journal ArticleDOI
TL;DR: Realtime Cerebellum (RC) is introduced, a new implementation of the large-scale spiking network model of the cerebellum which was originally built to study cerebellar mechanisms for simultaneous gain and timing control and acted as a general-purpose supervised learning machine of spatiotemporal information, on a graphics processing unit (GPU).

Journal ArticleDOI
TL;DR: This work considers individual, physical, and psychophysical factors that contribute to social spacing and demonstrates the feasibility of autonomous real-time annotation of these proxemic features during a social interaction between two people and a humanoid robot in the presence of a visual obstruction.
Abstract: In this work, we discuss a set of feature representations for analyzing human spatial behavior (proxemics) motivated by metrics used in the social sciences. Specifically, we consider individual, physical, and psychophysical factors that contribute to social spacing. We demonstrate the feasibility of autonomous real-time annotation of these proxemic features during a social interaction between two people and a humanoid robot in the presence of a visual obstruction (a physical barrier). We then use two different feature representations—physical and psychophysical—to train Hidden Markov Models (HMMs) to recognize spatiotemporal behaviors that signify transitions into (initiation) and out of (termination) a social interaction. We demonstrate that the HMMs trained on psychophysical features, which encode the sensory experience of each interacting agent, outperform those trained on physical features, which only encode spatial relationships. These results suggest a more powerful representation of proxemic behavior with particular implications in autonomous socially interactive and socially assistive robotics.

Proceedings ArticleDOI
01 Oct 2013
TL;DR: An approach to the acquisition of affordances and tool use in a humanoid robot combining vision, learning and control is described and learning is structured to enable a natural progression of episodes that include objects, tools, and eventually knowledge of the complete task.
Abstract: One of the recurring challenges in humanoid robotics is the development of learning mechanisms to predict the effects of certain actions on objects. It is paramount to predict the functional properties of an object from “afar”, for example on a table, in a rack or a shelf, which would allow the robot to select beforehand and automatically an appropriate action (or sequence of actions) in order to achieve a particular goal. Such sensory to motor schemas associated to objects, surfaces or other entities in the environment are called affordances [1, 2] and, more recently, they have been formalized computationally under the name of object-action complexes [3] (OACs). This paper describes an approach to the acquisition of affordances and tool use in a humanoid robot combining vision, learning and control. Learning is structured to enable a natural progression of episodes that include objects, tools, and eventually knowledge of the complete task. We finally test the robot's behavior in an object retrieval task where it has to choose among a number of possible elongated tools to reach the object of interest which is otherwise out of the workspace.

Proceedings ArticleDOI
24 Apr 2013
TL;DR: In this article, the authors present a complete, exact, analytical solution to both forward and inverse kinematics for the Aldebaran NAO humanoid robot and present a software library implementation for real-time onboard execution.
Abstract: The design of complex dynamic motions for humanoid robots is achievable only through the use of robot kinematics. In this paper, we study the problems of forward and inverse kinematics for the Aldebaran NAO humanoid robot and present a complete, exact, analytical solution to both problems, including a software library implementation for realtime onboard execution. The forward kinematics allow NAO developers to map any configuration of the robot from its own joint space to the three-dimensional physical space, whereas the inverse kinematics provide closed-form solutions to finding joint configurations that drive the end effectors of the robot to desired target positions in the three-dimensional physical space. The proposed solution was made feasible through a decomposition into five independent problems (head, two arms, two legs), the use of the Denavit-Hartenberg method, and the analytical solution of a non-linear system of equations. The main advantage of the proposed inverse kinematics solution compared to existing approaches is its accuracy, its efficiency, and the elimination of singularities. In addition, we suggest a generic guideline for solving the inverse kinematics problem for other humanoid robots. The implemented, freely-available, NAO kinematics library, which additionally offers center-of-mass calculations, is demonstrated in two motion design tasks: basic center-of-mass balancing and pointing to the ball.

Journal ArticleDOI
TL;DR: A syntactic approach to imitation learning that captures important task structures in the form of probabilistic activity grammars from a reasonably small number of samples under noisy conditions and can be recursively applied to help recognize unforeseen, more complicated tasks that share underlying structures.