“Hyper-adaptability” is the capability of the central nervous system (brain and spinal cord) to manage the impairment of sensory, motor, and cognitive functions by reactivating and recruiting pre-existing networks that are latent but available. In studying hyper-adaptability, interdisciplinary research that integrates mathematical modeling and robotic technologies with neuroscience findings is important. Thanks to support from Japan’s Ministry of Education, Culture, Sports, Science and Technology (MEXT) in the form of Grant-in-Aid for Scientific Research on Innovative Areas, the Hyper-adaptability Research Project started in July 2019.

The research works reported in this special issue are the latest achievements of the Hyper-adaptability Research Project. The special issue consists of three review articles and 10 research papers. These contributions cover a wide range of hyper-adaptability research activities, including neurophysiological experiments, functional neuroimaging, mathematical modeling of brain and musculoskeletal systems, cognitive psychological experiments, and robotic / virtual reality interventions for neurorehabilitation, carried out to clarify the mechanisms of hyper-adaptability.

We thank the authors for submitting their latest achievements and the reviewers for dedicating their time and effort to the review process. We also thank the editorial board of the Journal of Robotics and Mechatronics for supporting the publication of this special issue.

Although brain damage causes functional impairment, it is often followed by partial or total recovery of function. Recovery is believed to occur primarily because of brain plasticity. Both human and animal studies have significantly contributed to uncovering the neuronal basis of plasticity. Recent advances in brain imaging technology have enabled the investigation of plastic changes in living human brains. In addition, animal experiments have revealed detailed changes at the neural and genetic levels. In this review, plasticity in motor-related areas of the cerebral cortex, which is one of the most well-studied areas of the neocortex in terms of plasticity, is reviewed. In addition, the potential of technological interventions to enhance plasticity and promote functional recovery following brain damage is discussed. Novel neurorehabilitation technologies are expected to be established based on the emerging research on plasticity from the last several decades.

Allostasis is a physiological principle based on a dynamic regulatory system, contrary to homeostasis, in which the goal is to reach a steady state and recover from deviation from a set point in the internal environment. The concept of allostasis has continued to develop with advances in the field of neuroscience. In this short review, the author presents several new findings in neuroscience and extend the concept of allostasis as mutual regulation between cognitive, somatic, and autonomic systems. In this manner, biological systems adapt to external and internal environments by changing themselves.

Many stroke patients suffer from motor impairments due to paralysis, and consequently, motor paralysis of upper limbs seems to be particularly prone to residual impairment compared to that of lower limbs. Although ‘learned non-use’ that by managing reasonably well using only the unaffected upper limb in their actions, the patients can achieve their desired behavior, and these success experiences strengthen this pattern of behavior can be interpreted as a post-stroke adaptation, physiotherapy may lead to poor recovery of motor impairment. This review article discusses the impact of upper limb physiotherapy after stroke on behavioral/brain adaptations. Our previous studies demonstrated that patients with severe post-stroke sensorimotor impairments in a chronic phase might have abnormal functional connectivity. To prevent such adaptation after stroke, upper limb physiotherapy is important. In rehabilitation practices, hyper-adaptation has been often observed in not only behavioral but also brain changes. Although several studies are reporting clinical efficacy in patients with moderate to mild paralysis, there might be no effective treatment for patients with severe motor paralysis. To overcome these serious problems, we have developed a novel approach, kinesthetic illusion induced by visual stimulation (KINVIS) therapy. We showed that the effects of KINVIS therapy with therapeutic exercise on upper limb motor functions were mediated by spasticity, and functional connectivity in the brain was also changed with the improvement of motor function after KINVIS therapy. Brain changes underlying behavioral changes need to be more examined, and the adaptation of stroke patients needs to be clarified in detail.

Severe and multiple disabilities (SMD) refers to the simultaneous occurrence of intellectual and physical problems. SMD in children is difficult to assess, as they often do not have the proper language or bodily responses to represent their feelings. In this study, we propose a methodology for evaluating reactions of children with SMD to sensory stimuli that does not rely on observations by humans, but rather is based on automatic detection of video-recorded data and quantification by time-series analyses. We present two case studies with typical participants: one with large body movements (P1) and another with subtle body movements (P2). For P1, it was observed that he showed larger bodily movements just before the onset of tactile stimuli, while he became silent for approximately 10 s after the onset, with the stimuli causing him to reduce self-stimulatory behavior and pay attention to his external environment. For P2, two quantitative methodologies – correlation coefficient and Granger causality – were adopted, to compare behavioral difference during the presentation of either sour or sweet taste stimuli. For the sweet conditions, the movement of the mouth was considered to be generated by some internal causes. Through these experiments, we confirmed the authenticity of assessments made by the participants’ caregivers, and also revealed otherwise unseen behavioral patterns and structures.

Regional cerebral activity related to attention may be more useful as an evaluation index for attention levels than conventional task performance score-based methods. We therefore researched whether the quantitative evaluation of attention using regional cerebral activity, measured using near-infrared spectroscopy (NIRS), was appropriate. NIRS signals during the continuous performance test (CPT), which is well known as an attention test, were measured and analyzed. We confirmed activities in the regions that may be associated with the right-side anterior cingulate cortex (ACC), and on the estimated dorsolateral prefrontal cortex (DLPFC). Furthermore, there was a high correlation between activity on the DLPFC related to executive function and the performance score. Our study using cerebral activity could not quantify attention, but it opened the possibility of quantifying levels of executive function.

We investigated the ability of humans to adapt to a novel environment by kinematic transformation. This adaptation was studied via behavioural experiments using a robotic manipulandum – a system designed to arbitrarily generate virtual force fields against a human hand and subsequently record the hand’s trajectory. By repeating motor tasks, this study’s participants gradually learned to move correctly under a newly experienced force field, such as rotating in a clockwise direction. However, each participant’s motor memory was destroyed if he/she experienced an opposing force field (e.g., in a counterclockwise direction) immediately after learning the initial movement, which is known as retrograde interference. In some previous studies, it has been considered that by presenting sensory cues to highlight the difference in two opposing force fields, participants can learn both force fields independently without interference. In this study, we investigated the functionality of olfactory cues – specifically lemon and lavender odors – in reducing retrograde interference. Forty-five university students participated in an experiment using a robotic manipulandum. Our results have shown that the presence of lemon odor reduces the destruction of motor memory, while that of lavender did not, suggesting that odors can enhance simultaneous motor learning but the effect depends on the type of odor used.

Among several perceptive traits of virtual reality, the relationship between the physical body and a self-avatar is unclear. In this study, we investigate a case of hyper-adaptability, i.e., the capability of users to adjust to the movements of an altered self-avatar when such movements abruptly and frequently change. Focusing on movements of the upper limbs, we show experimentally the effect of the frequency of variations in virtual body alterations on adaptability. Moreover, we report a positive evaluation of the sense of embodiment and the overall user experience with virtual reality, and finally underline how these studies can be considered a basis for the design and development of virtual rehabilitation systems.

Sit-to-stand motion is an important daily activity, and disability of motion can significantly reduce quality of life. Therefore, it is important to understand the mechanism of sit-to-stand motion in order to prevent such scenarios. The sit-to-stand motion was found to be generated by four muscle groups, through muscle synergy. However, it is unclear how muscle synergy can be controlled. Human sit-to-stand motion may be planned based on body condition before motion. In this study, we aimed to clarify the relationship between body condition and muscle activity during the sit-to-stand motion. Accordingly, we measured the muscle activity when knee movability was disturbed as a condition of body change. We also measured the muscle activity during normal sit-to-stand motion and sit-to-stand motion with disturbed knee movability using surface electromyography. Subsequently, we extracted the muscle synergy from the measured muscle activity and compared the activity levels of muscle synergy. The results revealed that muscle activity contributing to forward bending increased and that contributing to the rise of the hip and stabilization decreased when knee movability was disturbed. These results suggest that humans compensate for disturbed knee movability with forward momentum and bending motion. Moreover, this implies that humans adjust their motion to various environments or body conditions by adjusting the level of forward bending activity.

Robot-assisted body movements are a useful approach for the rehabilitation of motor dysfunction. Various robots based on end-effector or exoskeleton type have been proposed. However, the effect of these robots on brain activity during assistive lower limb movements remains unclear. In this study, we evaluated brain activity results among robot-assisted passive movements, voluntary active movements, and kinesthetic motor imagery. We measured and compared the brain activities of 21 young, healthy individuals during three experimental conditions associated with lower limb movements (active, passive, and imagery conditions) using functional near-infrared spectroscopy (fNIRS). Our results showed that although different brain areas with significant activity were observed among the conditions, the temporal patterns of the activity in each recording channel and the spatial patterns of functional connectivity showed high similarity between robot-assisted passive movements and voluntary active movements. Conversely, the robot-assisted passive movements did not show any similarity to motor imagery. Overall, these findings suggest that the robotic assistive approach is useful for activating not only afferent processes associated with sensory feedback processing but also motor control-related efferent processes.

Human motor learning is characterized by adaptation, wherein information obtained in the past is transferred to a different situation. In this study, we investigate a grid-based computation for explaining the reuse of the information of an existing controller for adaptation to a partial malfunction of a controller. To this end, a motor learning scheme is adopted based on the detection and estimation of partial relationships. The transformation between the partial relationships is estimated based on a grid-based estimation of the two coordinate systems. In this estimation, the coordinate systems are optimized using a genetic algorithm. Two arms in a reflection are considered, and it is confirmed that the transformation of the differential kinematics (Jacobian), as an example of the partial relationships, can be estimated by the proposed method.

The optic lobe of the fly is one of the prominent model systems for the neural mechanism of the motion detection. How a fly who lives under various visual situations of the nature processes the information from at most a few thousands of ommatidia in their neural circuit for the detection of moving objects is not exactly clear though many computational models of the fly optic lobe as a moving objects detector were suggested. Here we attempted to elucidate the mechanisms of ON-edge motion detection by a simulation approach based on the TEM connectome of Drosophila. Our simulation model of the optic lobe with the NEURON simulator that covers the full scale of ommatidia, reproduced the characteristics of the receptor neurons, lamina monopolar neurons, and T4 cells in the lobula. The contribution of each neuron can be estimated by changing synaptic connection strengths in the simulation and measuring the response to the motion stimulus. Those show the paradelle pathway provide motion detection in the fly optic lobe has more robustness and is more sophisticated than a simple combination of HR and BL systems.

People and animals adapt their gait to the environment as they perform activities in a variety of environments. However, there are cases where the parts of the body necessary for walking are damaged in some way, resulting in walking difficulties. An example is paralysis caused by a stroke. A split-belt treadmill is occasionally used for the investigation to analyze how the stroke effects on the motion. However, the mechanical properties of the split-belt treadmill on the body have not been clarified. It is also unknown how the mechanical closed-loop between the body and the environment, generated by synchronizing the movements of the two belts, affects the gait. In this study, we investigated that the effect of the mechanical closed-loop structure between the body and the environment on walking using the robot and the mechanical effect of the floor reaction force on the body. Further, we conducted walking experiments using the developed robot, obtained body and environmental information, and analyzed the results. As the result, it was observed that the motion data differed based on the coupling of the treadmill. In other words, it was suggested that the mechanical closed-loop structure certainly influenced the physical balances on walking motion. Furthermore, it is confirmed that the coupling of treadmills increases the body’s sway. Although our results are given from a robotic experiment, it is expected that these measures would be one of the important index in human rehabilitations.

When considering the human motor-adaptation mechanism from the perspective of the motor control theory, updating the internal model constitutes a critical component. The learning curve at each trial of motion can be explained by a state-space model; however, the model cannot reproduce the time-series data for the hand’s position, velocity, and acceleration (motion profiles). There is no internal model-updating rule for optimal feedback control, a plausible model for reproducing motion profiles. In this paper, we propose an adaptation model that incorporates an internal model-updating rule which modeled after Hebb’s rule into optimal feedback control. Also, we examine the neural substrate of the internal model. To validate the proposed adaptation model, we conducted behavioral experiments with humans that reflected changes in the internal model and reproduced the changes in the internal model as well as the motion profiles using the proposed adaptation model. In addition, we analyzed the data for a visuomotor rotation task performed by a monkey and checked for changes in the output characteristics of neurons in the motor cortex before and after adaptation. According to the above-mentioned validation and analysis results, the motor cortex constitutes the neural substrate of the internal model.

In the past decades, robot navigation in an unknown environment has attracted extensive interest due to its tremendous application potential. However, most existing schemes rely on complex sensing systems and control systems to perceive and process the geometric and appearance information of the surrounding environment to avoid the collision, while making less use of the mechanical characteristics of the environment. In this research, in order to explore how to make a robot navigate in an unknown environment with minimal active control and minimal sensing by taking full advantage of the mechanical interactions from the environment, which is called implicit control in this study, we propose a centipede robot and its corresponding navigation scheme for navigating a 2D unknown environment without sensing information about the surrounding environment. In this scheme, the only observation input of this system is the goal direction information relative to the robot direction. Based on this scheme, we built a prototype robot and conducted navigation experiments in three environments with different levels of complexity. As a result, we obtained the navigation route map and navigation time distribution of each environment and analyzed the characteristics and applicability scenarios of the proposed navigation scheme compared to the traditional ones.

Surgical training simulators with virtual reality have been developed to enable surgeons to efficiently acquire and improve their surgical skills. In hard tissue surgery, the surgeon uses a chisel and mallet to cut a bone or tooth with large and instantaneous forces. In the previous study by present authors, to represent the force sensation of the cutting operation in the virtual training simulator, we constructed the force display device using the ball-screw mechanism to obtain high stiffness and display the large force. Additionally, we applied the two-degrees-of-freedom (2DOF) admittance control to react instantaneously to the impact force by pounding with the mallet. The feedback controller of the 2DOF admittance control is required to increase the high-frequency gain for improving the responsiveness of the force display device. However, the vibrational mode of the force display device can be excited by increasing the controller gain. Therefore, this study develops the design approach of the feedback controller using the H_∞ control in the 2DOF admittance control system, which can be systematically constructed to reduce the vibrational mode and react instantaneously in the force display device. The efficacy of the proposed force display control system is verified through the virtual experience of the free movement and the hard contact operations.

In our previous study, we developed a walking support shoe with an elastomer-embedded flexible joint (EEFJ) to assist the function of tibialis anterior (TA) in initial stances (IC) and swing phases (SW). However, its usability and supporting effect have not been sufficiently evaluated. Therefore, in this study, we developed a dorsiflexion support unit (DSU) using the EEFJs with consideration on the usability for frail persons. Their needs were investigated in hearings at community centers. With reference to their comments, we proposed a three-phased scenario in which pre- / post-activities were considered as important factors of its product design of the DSU. We designed the DSU for better usability in the pre- / post-activities. Its basic function and mechanical properties were also investigated in experiments. According to the mechanical tests, the supporting torque was around 10% of the activation of TA in IC. In addition, the results of gait tests show reductions of ankle rotations by 17% and 11% in IC and SW, respectively, without significant increases of TA activations.

For autonomous navigation, we have thus far proposed MCL using environmental maps based on cadastral data. However, buildings in the cadastral data sometimes differ from their actual positions in the environment. As the environmental map is generated from the cadastral data, the inconsistency affects the localization performance. For this problem, we propose an online SLAM approach in the actual environment. A mobile robot simultaneously localizes the position and builds another online map using NDT scan matching. In contrast to other offline SLAM approaches, however, pose graph optimization for loop closure is not executed during online SLAM. As a result, the online map is distorted by localization errors. For this challenge inherent in online SLAM, the localization errors are modified using MCL and wheel odometry in a hybrid manner. As a contribution to autonomous navigation, the robot is enabled to localize the position even in a new place. In the experiments, we show that the localization performance of the robot in an outdoor environment with inconsistent buildings is improved compared to other online approaches with and without modifications.

This paper presents a localization method using deep learning and light detection and ranging (LiDAR) for unmanned ground vehicle (UGV) in field environment. We develop a sensor fusion algorithm that UGV recognizes natural objects from RGB camera using deep learning and measures the distance to the recognized objects with LiDAR. UGV calculates its position relative to the objects, creates a reference path, and then executes path following control. The method is applied to autonomous mowing operation in orchard. A mower is tracked by UGV. UGV needs to run along a tree row keeping an appropriate distance from tree trunks, by which the mowing arm of the tracked mower properly touches the trunks. Field experiments are conducted in pear and apple orchards. UGV localizes self position relative to trees and performs autonomous mowing successfully. The results show that the presented method is effective.

Navigation requires self-localization, which GPS typically performs outdoors. However, GPS frequently introduces significant errors in environments where radio reflection occurs, such as in urban areas, which makes precise self-localization occasionally impossible. Meanwhile, a human can use street view images to understand their surroundings and their current location. To implement this, we must solve image matching between the current scene and the images in a street view database. However, because of the wide variations in the field angle, time, and season between images, standard pattern matching by feature is difficult. DeepMatching can precisely match images with various lighting and field angles. Nevertheless, DeepMatching tends to misinterpret street images because it may find unnecessary feature points in the road and sky. This study proposes several methods to reduce misjudgment: (1) gaining image similarity with features such as buildings by excluding the road and sky, and (2) splitting the panoramic image into four directions and matching in each. This study shows the results of each method and summarizes their performance using various images and resolutions.