Full-body Humanoid Robot Research Articles

HumanoidBot: framework for integrating full-body humanoid robot with open-domain chat system

ABSTRACT State-of-the-art open-domain dialogue systems have made significant progress over the past few years using new model architectures that reliably preserve the context of dialogue and large dataset collection. However, there are no examples of such state-of-the-art open-domain dialogue systems implemented on humanoid robots. Implementing a dialogue system that can naturally interact with humans is a fundamental issue influencing the future of humanoid robots. Therefore, this paper proposes HumanoidBot, a framework for implementing an open-domain chat system on a full-body humanoid robot. Furthermore, an evaluation experiment using a humanoid robot implemented with HumanoidBot verified how balance control affects the dialogue and impression of the robot in humanoid robot dialogue.

Active Balance Control of Humanoid Locomotion Based on Foot Position Compensation

A foot positioning compensator is developed in this paper for a full-body humanoid to retrieve its balance during continuous walking. An online Foot Position Compensator (FPC) is designed to improve the robustness of biped walking, which can modify predefined step position and step duration online with sensory feedback. Foot placement parameters are learned by the FPC based on the Policy Gradient Reinforcement Learning (PGRL) method. Moreover, the FPC assists the humanoid robot in rejecting external disturbances and recovering the walking position by re-planning the trajectories of walking pattern and the Center of Mass (CoM). An upper body pose control strategy is also presented to further enhance the performance of humanoid robots to overcome strong external disturbances. The advantages of this proposed method are that it neither requires prior information about the walking terrain conditions, nor relies on range sensor information for surface topology measurement. The effectiveness of the proposed method is verified via Webots simulation and real experiments on a full-body humanoid NAO robot.

Dynamic walking control of humanoid robots combining linear inverted pendulum mode with parameter optimization

To improve the robustness of biped walking, a model parameters optimization method based on policy gradient decent learning is presented. For the linear inverted pendulum mode-based model parameters optimization, firstly, select the input parameters of the inverted pendulum model and the torso attitude parameters of the robot as the correction variables and establish the correction equation. Then, using the tracking errors of center of mass (CoM) of the robot and the errors of the robot posture relative to the upright state of the body to establish the fitness function. According to the fitness function, the gain coefficients in the model parameters correction equation are optimized by using the strategy gradient learning method, and the modified gain parameters are substituted into the model parameters correction equation to obtain the correction amount. By applying the model parameters optimization strategy, the robot can quickly and in real time adjust the body posture and walking patterns under unknown disturbances, hence, the walking robustness can be enhanced. Simulation and experiments on a full-body humanoid robot NAO validate the effectiveness of the proposed method. The experiments show that the optimized model yields a more controlled, robust walk on NAO robot and on various surfaces without additional manual parameters tuning.

Active balance of humanoid movement based on dynamic task-prior system

To maintain human-like active balance in a humanoid robot, this article proposes a dynamic priority-based multitask algorithm to avoid self-collision during highly complex robotic whole-body motions and rebalance after external disturbance using momentum compensation strategies. On the one hand, the conflict between self-collision and self-balance constraints in task-space merging with end-effector tracking tasks are considered in the multitask algorithm to improve the robot’s balance. On the other hand, for self-balance during the robot’s movement, momentum compensation is considered as one task and utilized to reject unknown disturbances. Two strategies are put forward to restrain the sudden change in momentum. One is to calculate the correction in the joint space with the resolved momentum control (RMC) method; the other is to add an end-effector tracking motion in the task space. Simulations and experiments on a full-body humanoid robot validate the effectiveness of the proposed method. Experiments on both a simulation and a full-body humanoid are designed to validate the task-prior algorithm. With the proposed method, the humanoid robot succeeds in avoiding self-collision during movement and is able to rectify itself to the preplanned steady stance while encountering undefined external forces.

Dynamic and Reactive Walking for Humanoid Robots Based on Foot Placement Control

This paper presents a novel online walking control that replans the gait pattern based on our proposed foot placement control using the actual center of mass (COM) state feedback. The analytic solution of foot placement is formulated based on the linear inverted pendulum model (LIPM) to recover the walking velocity and to reject external disturbances. The foot placement control predicts where and when to place the foothold in order to modulate the gait given the desired gait parameters. The zero moment point (ZMP) references and foot trajectories are replanned online according to the updated foothold prediction. Hence, only desired gait parameters are required instead of predefined or fixed gait patterns. Given the new ZMP references, the extended prediction self-adaptive control (EPSAC) approach to model predictive control (MPC) is used to minimize the ZMP response errors considering the acceleration constraints. Furthermore, to ensure smooth gait transitions, the conditions for the gait initiation and termination are also presented. The effectiveness of the presented gait control is validated by extensive disturbance rejection studies ranging from single mass simulation to a full body humanoid robot COMAN in a physics based simulator. The versatility is demonstrated by the control of reactive gaits as well as reactive stepping from standing posture. We present the data of the applied disturbances, the prediction of sagittal/lateral foot placements, the replanning of the foot/ZMP trajectories, and the COM responses.

Realization and swimming performance of the breaststroke by a swimming humanoid robot

In order to clarify the mechanics of human swimming, a full-body swimming humanoid robot called “SWUMANOID” was developed as an experimental platform for research about human swimming. SWUMANOID had a detailed human body shape, created using three-dimensional scanning and printing equipment, and was developed as an experimental model substituting for human subjects. Not only the appearance but also the methodology to realize various swimming strokes was considered. In order to reproduce complicated swimming motions with high fidelity, 20 waterproof actuators were installed. The free swimming of the crawl stroke at a velocity of 0.24 m/s was realized in the previous study. However, it could not perform the breaststroke due to mechanical limitations. The objectives of this study were to realize the breaststroke for SWUMANOID by improving its lower limbs, and to investigate the swimming performance of the breaststroke experimentally. The lower body of SWUMANOID was fully redesigned, built, and connected to the upper body. The swimming motion of the breaststroke was created based on that of an actual swimmer. A free swimming experiment was conducted in a 25 m outdoor swimming pool. In addition, in order to discuss the experimental results in detail, the experiment was reproduced by the simulation. From the experiment, it was found that SWUMANOID could perform the breaststroke successfully. The swimming speed for the stroke cycle of 2.3 s was found to be 0.12 m/s. Since this swimming speed was considered low compared to that of the actual swimmer, the reason for the discrepancy was examined by simulation. From the simulation, it was found that one main reason for the low swimming speed was insufficient output power of the motors, especially for the knee and shoulder joints.

Active balance of humanoids with foot positioning compensation and non-parametric adaptation

To maintain human-like active balance for a humanoid robot, this paper proposes a novel adaptive non-parametric foot positioning compensation approach that can modify predefined step position and step duration online with sensor feedback. A constrained inverted pendulum model taking into account of supporting area to CoM acceleration is used to generate offline training samples with constrained nonlinear optimization programming. To speed up real-time computation and make online model adjustable, a non-parametric regression model based on extended Gaussian Process model is applied for online foot positioning compensation. In addition, a real-time and sample-efficient local adaptation algorithm is proposed for the non-parametric model to enable online adaptation of foot positioning compensation on a humanoid system. Simulation and experiments on a full-body humanoid robot validate the effectiveness of the proposed method.

휴머노이드 로봇 모델을 이용한 보행재활 훈련장치의 견인모터 속도 파형 생성

This work proposes a new method to generate velocity profile of a traction motor equipped in a rehabilitation system for knee joint patients through humanoid robot simulation. To this end, a three-dimensional full-body humanoid robot model is newly constructed, and natural human gait is simulated by applying to it reference joint angle trajectories already published. Linear velocity is derived from distance data calculated between the positions of a thigh band and its traction motor at every sampling instance, which is a novel idea of this paper. The projection rule is employed to kinematically describe the humanoid robot because of its high efficiency and accuracy, and measured joint trajectories are used in simulating human natural gait referring to Winter`s book. The attained motor velocity profile for a certain position in human body will be applied to our walking-assistance system which is implemented with a treadmill system.

Bipedal robot walking control on inclined planes by fuzzy reference trajectory modification

The two-legged humanoid structure has advantages for an assistive robot in the human living and working environment. However, the control of bipedal walk is a challenge. Walking performance on solely even floor is not satisfactory. This paper presents a study on bipedal walk on inclined planes with changing slopes. A Zero Moment Point (ZMP) based gait synthesis technique is employed. The pitch angle reference for the foot sole plane--as expressed in a coordinate frame attached at the robot body--is adjusted online by a fuzzy logic system to adapt to different walking surface slopes. Ankle pitch torques and the average value of the body pitch angle, computed over a history of a predetermined number of sampling instants, are used as the inputs to this system. The proposed control method is tested via walking experiments with the 29 degrees-of-freedom human-sized full-body humanoid robot SURALP (Sabanci University Robotics Research Laboratory Platform) on even floor and inclined planes with different slopes. The results indicate that the approach presented is successful in enabling the robot to stably enter, ascend and leave inclined planes with 15 % (8.5°) grade. This, to the best knowledge of the authors, constitutes the steepest ascend reported to date, with a transition from even floor, by a blind walking biped robot.

A Hybrid CPG–ZMP Controller for the Real-Time Balance of a Simulated Flexible Spine Humanoid Robot

Although a few researchers have started to realize the importance of a flexible spine in generating more natural-looking behaviors for humanoid robots, there has been no work on trying to make a full-body humanoid robot (that has a flexible spine) to maintain balance. The main reason for this is that it is very costly and difficult to develop and control this kind of robot. This paper serves two purposes. First, it proposes the use of a realistic 3-D robot simulator as a platform for costly flexible spine humanoid robotics research. Second, it presents a hybrid CPG-ZMP controller for the simulated robot. The biologically inspired CPG component of our controller allows the mechanical spine and feet to exhibit rhythmic motions using only two control parameters. Through monitoring the measured ZMP location, the engineering component modulates the neural activity of the CPG to allow the robot to maintain balance while it is standing and exhibiting motions on the sagittal and frontal planes in real time. The final postures of the simulated humanoid emerge automatically in real time through dynamic interactions between the neural networks, the robot itself, and its environment. Since experimental results have also demonstrated that our system is robust against disturbances from external pushing forces, our controller has the potential to be applicable to the next generation of humanoid robots.

EFFECT OF A FLEXIBLE SPINE EMOTIONAL BELLY DANCING ROBOT ON HUMAN PERCEPTIONS

Recently, there has been a growing interest in human–robot interaction. Researchers in artificial intelligence and robotics have built various types of social robots which can express emotions through speech, facial expressions and hand gestures. Although some of these robots are able to interact with humans in interesting ways, they cannot move as naturally as we do because of the limited number of degrees of freedom in their body torsos (some of them do not even have a torso). Since we often express and perceive each other's emotions and motives at a distance using body language alone, it would be good for the next generation of humanoid robots to possess similar capabilities. As a first step towards this goal, we developed a 28-DOF full-body humanoid robot as an experimental platform. Unlike the current generation of humanoid robots, our robot has a flexible spine. This feature is very important because counterbalancing movements of the spine are required to maintain dynamic stability in humans and humanoid robots. Our robot can belly dance and communicate affective motions via full-body movements. Using a Central Pattern Generator (CPG) based controller, we generated rhythmic motions for the arms, upper and lower bodies. We then conducted psychological experiments using both the robot and a human actors. Statistical analyses were carried out to test our hypotheses on human perception of affective movements. Experimental results show that human subjects were able to perceive emotions from the robot based only on its body motions, sometimes as well as recognizing the movements being performed by the human actor. Our robot can be used to examine the relationship between the movement of the spine, shoulders, arms, neck and head when attempting to reproduce affective movements. Psychologists, actors, dancers and animators can benefit from this line of research by learning how emotions can be conveyed through body motions and knowing how body part movements combine to communicate emotional expressions.

Discovering optimal imitation strategies

This paper develops a general policy for learning the relevant features of an imitation task. We restrict our study to imitation of manipulative tasks or gestures. The imitation process is modeled as a hierarchical optimization system, which minimizes the discrepancy between two multi-dimensional datasets. To classify across manipulation strategies, we apply a probabilistic analysis to data in Cartesian and joint spaces. We determine a general metric that optimizes the policy of task reproduction, following strategy determination. The model successfully discovers strategies in six different imitative tasks and controls task reproduction by a full body humanoid robot.

Discovering optimal imitation strategies

This paper develops a general policy for learning the relevant features of an imitation task. We restrict our study to imitation of manipulative tasks or gestures. The imitation process is modeled as a hierarchical optimization system, which minimizes the discrepancy between two multi-dimensional datasets. To classify across manipulation strategies, we apply a probabilistic analysis to data in Cartesian and joint spaces. We determine a general metric that optimizes the policy of task reproduction, following strategy determination. The model successfully discovers strategies in six different imitative tasks and controls task reproduction by a full body humanoid robot.