Search Results (8)

In imitation learning with the embodiment gap, directly transferring human motions to robots is challenging due to differences in body structures. Therefore, it is necessary to reconstruct human motions in accordance with each robot’s embodiment. Our previous work focused on the right arm of a humanoid robot, which limited the generality of the approach. To address this, we propose Cross-Embodiment Kinematic Behavioral Cloning (X-EKBC), an imitation learning framework that enables movement-level imitation on a one-to-one basis between humans and multiple robots with embodiment gaps. We introduce a joint matrix that represents the structural correspondence between the human and robot bodies, and by solving kinematics based on this matrix, the system can efficiently reconstruct motions adapted to each robot’s embodiment. Furthermore, by employing Implicit Behavioral Cloning (IBC), the proposed method achieves both imitation learning of the reconstructed motions and quantitative evaluation of embodiment gaps using energy-based modeling. As a result, motion reconstruction through the joint matrix became feasible, enabling both imitation learning and quantitative embodiment evaluation based on reconstructed behaviors. Future work will aim to extend this framework toward motion-level imitation that captures higher-level behavioral outcomes. Full article

This paper proposes a deep reinforcement learning-based predictive control scheme to address cushion pressure prediction and stabilization in hovercraft systems subject to modeling complexity, dynamic instability, and system delay. Notably, this work introduces a long short-term memory (LSTM) network with a temporal sliding window specifically designed for hovercraft cushion pressure forecasting. The model accurately captures the dynamic coupling between fan speed and chamber pressure while explicitly incorporating inherent control lag during airflow transmission. Furthermore, a novel adaptive behavior cloning mechanism is embedded into the twin delayed deep deterministic policy gradient with behavior cloning (TD3-BC) framework, which dynamically balances reinforcement learning (RL) objectives and historical policy constraints through an auto-adjusted weighting coefficient. This design effectively mitigates distribution shift and policy degradation in offline reinforcement learning, ensuring both training stability and performance beyond the behavior policy. By integrating the LSTM prediction model with the adaptive TD3-BC algorithm, a fully data-driven control architecture is established. Finally, simulation results demonstrate that the proposed method achieves high accuracy in cushion pressure tracking, significantly improves motion stability, and extends the operational lifespan of lift fans by reducing rotational speed fluctuations. Full article

This study investigates how the behavioural realism of avatars can enhance virtual reality (VR) social interactions involving self-disclosure. First, we review how factors such as trust, enjoyment, and nonverbal communication could be influenced by motion capture technology by enhancing behavioural realism. We also address a gap in the prior literature by comparing different motion capture systems and how these differences affect perceptions of realism, enjoyment, and eye contact. Specifically, this study compared two types of avatars: an iClone UNREAL avatar with full-body and facial motion capture and a Vive Sync avatar with limited motion capture for self-disclosure. Our participants rated the iClone UNREAL avatar higher for realism, enjoyment, and eye contact duration. However, as shown in our post-experiment survey, some participants reported that they preferred the avatar with less behavioural realism. We conclude that a higher level of behavioural realism achieved through more advanced motion capture can improve the experience of VR social interactions. We also conclude that despite the general advantages of higher motion capture, the simpler avatar was still acceptable and preferred by some participants. This has important implications for improving the accessibility of avatars for different contexts, such as therapy, where simpler avatars may be sufficient. Full article

This retrospective begins with Galvani’s experiments on frogs at the end of the 18th century and his discovery of ‘animal electricity’. It goes on to illustrate the numerous contributions to the field of physical chemistry in the second half of the 19th century (Nernst’s equilibrium potential, based on the work of Wilhelm Ostwald, Max Planck’s ion electrodiffusion, Einstein’s studies of Brownian motion) which led Bernstein to propose his membrane theory in the early 1900s as an explanation of Galvani’s findings and cell excitability. These processes were fully elucidated by Hodgkin and Huxley in 1952 who detailed the ionic basis of resting and action potentials, but without addressing the question of where these ions passed. The emerging question of the existence of ion channels, widely debated over the next two decades, was finally accepted and, a decade later, many of them began to be cloned. This led to the possibility of modelling the activity of individual neurons in the brain and then that of simple circuits. Taking advantage of the remarkable advances in computer science in the new millennium, together with a much deeper understanding of brain architecture, more ambitious scientific goals were dreamed of to understand the brain and how it works. The retrospective concludes by reviewing the main efforts in this direction, namely the construction of a digital brain, an in silico copy of the brain that would run on supercomputers and behave just like a real brain. Full article

The lack of intuitive controllability remains a primary challenge in enabling transhumeral amputees to control a prosthesis for arm reaching with residual limb kinematics. Recent advancements in prosthetic arm control have focused on leveraging the predictive capabilities of artificial neural networks (ANNs) to automate elbow joint motion and wrist pronation–supination during target reaching tasks. However, large quantities of human motion data collected from different subjects for various activities of daily living (ADL) tasks are required to train these ANNs. For example, the reaching motion can be altered when the height of the desk is changed; however, it is cumbersome to conduct human experiments for all conditions. This paper proposes a framework for cloning motion datasets using deep reinforcement learning (DRL) to cater to training data requirements. DRL algorithms have been demonstrated to create human-like synergistic motion in humanoid agents to handle redundancy and optimize movements. In our study, we collected real motion data from six individuals performing multi-directional arm reaching tasks in the horizontal plane. We generated synthetic motion data that mimicked similar arm reaching tasks by utilizing a physics simulation and DRL-based arm manipulation. We then trained a CNN-LSTM network with different configurations of training motion data, including DRL, real, and hybrid datasets, to test the efficacy of the cloned motion data. The results of our evaluation showcase the effectiveness of the cloned motion data in training the ANN to predict natural elbow motion accurately across multiple subjects. Furthermore, motion data augmentation through combining real and cloned motion datasets has demonstrated the enhanced robustness of the ANN by supplementing and diversifying the limited training data. These findings have significant implications for creating synthetic dataset resources for various arm movements and fostering strategies for automatized prosthetic elbow motion. Full article

Bipedal robots have gained increasing attention for their human-like mobility which allows them to work in various human-scale environments. However, their inherent instability makes it difficult to control their balance while they are physically interacting with the environment. This study proposes a novel balance controller for bipedal robots based on a behavior cloning model as one of the machine learning techniques. The behavior cloning model employs two deep neural networks (DNNs) trained on human-operated balancing data, so that the trained model can predict the desired wrench required to maintain the balance of the bipedal robot. Based on the prediction of the desired wrench, the joint torques for both legs are calculated using robot dynamics. The performance of the developed balance controller was validated with a bipedal lower-body robotic system through simulation and experimental tests by providing random perturbations in the frontal plane. The developed balance controller demonstrated superior performance with respect to resistance to balance loss compared to the conventional balance control method, while generating a smoother balancing movement for the robot. Full article

Multiple myeloma (MM) is a genetically complex hematological cancer that is characterized by proliferation of malignant plasma cells in the bone marrow. MM evolves from the clonal premalignant disorder monoclonal gammopathy of unknown significance (MGUS) by sequential genetic changes involving many different genes, resulting in dysregulated growth of multiple clones of plasma cells. The migration, survival, and proliferation of these clones require the direct and indirect interactions with the non-hematopoietic cells of the bone marrow. We develop a hybrid discrete-continuous model of MM development from the MGUS stage. The discrete aspect of the modelisobservedatthecellularlevel: cellsarerepresentedasindividualobjectswhichmove,interact, divide, and die by apoptosis. Each of these actions is regulated by intracellular and extracellular processes as described by continuous models. The hybrid model consists of the following submodels that have been simplified from the much more complex state of evolving MM: cell motion due to chemotaxis, intracellular regulation of plasma cells, extracellular regulation in the bone marrow, and acquisition of mutations upon cell division. By extending a previous, simpler model in which the extracellular matrix was considered to be uniformly distributed, the new hybrid model provides a more accurate description in which cytokines are produced by the marrow microenvironment and consumed by the myeloma cells. The complex multiple genetic changes in MM cells and the numerous cell-cell and cytokine-mediated interactions between myeloma cells and their marrow microenviroment are simplified in the model such that four related but evolving MM clones can be studied as they compete for dominance in the setting of intraclonal heterogeneity. Full article

State estimation is the most critical capability for MAV (Micro-Aerial Vehicle) localization, autonomous obstacle avoidance, robust flight control and 3D environmental mapping. There are three main challenges for MAV state estimation: (1) it can deal with aggressive 6 DOF (Degree Of Freedom) motion; (2) it should be robust to intermittent GPS (Global Positioning System) (even GPS-denied) situations; (3) it should work well both for low- and high-altitude flight. In this paper, we present a state estimation technique by fusing long-range stereo visual odometry, GPS, barometric and IMU (Inertial Measurement Unit) measurements. The new estimation system has two main parts, a stochastic cloning EKF (Extended Kalman Filter) estimator that loosely fuses both absolute state measurements (GPS, barometer) and the relative state measurements (IMU, visual odometry), and is derived and discussed in detail. A long-range stereo visual odometry is proposed for high-altitude MAV odometry calculation by using both multi-view stereo triangulation and a multi-view stereo inverse depth filter. The odometry takes the EKF information (IMU integral) for robust camera pose tracking and image feature matching, and the stereo odometry output serves as the relative measurements for the update of the state estimation. Experimental results on a benchmark dataset and our real flight dataset show the effectiveness of the proposed state estimation system, especially for the aggressive, intermittent GPS and high-altitude MAV flight. Full article