Localization Task Research Articles

Precise visual positioning of agricultural mobile robots with a fiducial marker reprojection approach

Semi-structured greenhouse environment often features repetitive and weak texture, naturally bringing challenges to high-precision vision-based positioning techniques. This paper proposes a precise visual positioning method for agricultural mobile robots in the greenhouse, which improves their positioning accuracy via discriminatively minimizing fiducial marker reprojection errors. First, fiducial markers are used to enhance environment features, and a markers-based visual positioning task is formulated as a Perspective-n-Point (PnP) problem. The projection constraints of keypoints and the pose constraints of the coordinate systems provide a theoretical basis for robot positioning. Second, a reprojection error minimization approach is proposed by taking into account the markers’ distance and image noise. Far-away markers are more prone to greater observation errors than those close to the robots, the improved PnP algorithm considering distance weighting ensures higher positioning accuracy. Synthetic and field experiments are carried out to evaluate the performance of the proposed method. Synthetic experiments show that the rotation error and translation error of the proposed method are less than 0.7° and 0.5% within a range of 12 m. The mean absolute error and root mean square error of field dynamic positioning experiments are 8.57 cm and 8.59 cm, respectively. Experimental results show that the proposed method is significantly better than traditional methods in dealing with distance-related noise at keypoints.

An Exosome-Based Therapeutic Strategy Targeting Neuroinflammation in Alzheimer's Disease with Berberine and Palmatine.

Neuroinflammation is one of the major pathogeneses in Alzheimer's disease (AD) and mainly involves abnormal inflammatory activation of microglia by multiple pathological stimuli. The treatment of AD remains a major challenge due to the multifactorial characterization of AD and the inefficient ability of therapeutic drugs to permeate through the blood‒brain barrier (BBB). Accordingly, drug combination treatment and drug carrier delivery have become important therapeutic tools for the treatment of multifactorial diseases, especially AD. Inflammatory cytokine levels in microglia, including NO, TNF-α, IL-1β, IL-4, and IL-10, were detected. The Morris water maze and object location task were used to investigate the learning and memory functions of APP/PS1 mice in different treatment groups. The number of neurons and plasticity of synapses were evaluated by immunofluorescence double labelling. Additionally, the ratio of β-amyloid plaques and the number of activated microglia were evaluated by immunofluorescence staining. The concentrations of β-amyloid plaques and inflammatory factors in the hippocampus were determined by ELISA. Microglia-derived exosomes (Exos) were extracted and purified by size exclusion chromatography. The distribution of exosomes and drugs was investigated in vitro and in vivo. Compared to single drug interventions, the combination of Ber and Pal (Ber/Pal) modulated microglial inflammatory cytokine levels. Ber/Pal promoted the recovery of learning and memory impairment in APP/PS1 mice. Immunofluorescence staining indicated that Ber/Pal restored neurons, inhibited Aβ plaque formation and microglial activation, and regulated the secretion of inflammatory factors. Exos promoted the accumulation of drugs in cells and tissues and improved the targeting of drugs across the BBB. Ber/Pal could offer a synergistic and more comprehensive therapeutic effect in AD. Additionally, the microglia-derived Exos-Ber/Pal delivery system promoted the targeting and permeation of drugs into the brain, suggesting a creative strategy for targeting AD therapy by regulating neuroinflammation in microglial cells.

Transcranial direct current stimulation over dorsolateral prefrontal cortex improves postural stability in non-specific chronic low back pain patients with high fear of pain: A randomized sham-controlled trial.

Deficits in postural stability in response to environmental challenges are seen in chronic low back pain (CLBP) patients with high fear of pain (HFP). Hence, it is essential to follow effective approaches to treat postural stability deficits and improve the health status of these patients. The current study aimed to compare the effects of cathodal and anodal transcranial direct current stimulation (c-tDCS and a-tDCS) over the left dorsolateral prefrontal cortex (DLPFC) on postural stability in non-specific CLBP patients with HFP. In this randomized clinical trial study, 75 patients were randomly assigned to left DLPFC a-tDCS, left DLPFC c-tDCS and sham stimulation groups (n= 25 in each group). All groups received a single-session 2mA tDCS for 20 min, but the stimulation was slowly turned off after 30 s in the sham group. Before, immediately, 24 h and 1 week after the interventions, postural stability indices were assessed using a Biodex Balance System. A significant reduction in most indices was shown in both a-tDCS and c-tDCS groups after interventions (immediately, 24 h and 1 week follow-up) during static and dynamic postural tasks compared with the sham tDCS group (p< .01). In addition, some tests showed a significant difference between a-tDCS and c-tDCS (p< .05). The findings indicated positive effects of both a-tDCS and c-tDCS on the left DLPFC, with more efficacy of a-tDCS on postural stability in LBP patients with HFP.

Effects of expertise in skill-oriented sports on postural control: An event-related potential study using dual-task paradigm

Postural control is a fundamental aspect of motor skills and is crucial for achieving optimal performance in sports. However, previous studies mainly explored the postural control ability of athletes of skill-oriented sports (skilled athletes) under single postural control conditions, disregarding the processing characteristics of postural control in real sports scenarios. This research used the event-related potential (ERP) to investigate the postural control ability of skilled athletes during the action inhibition process by a dual-task paradigm. The study included 26 skilled athletes and 25 nonathletes who simultaneously completed a postural control task and a Go/Nogo task. The results showed that skilled athletes exhibited superior postural control performance in all three standing positions and higher N1 and P3 component amplitudes than nonathletes. The study illustrated that skilled athletes exhibit higher flexibility in allocating attentional resources to ensure optimal postural control and action inhibition performance. These findings suggest that skilled athletes possess a higher level of control automation in complex standing positions and more efficient and economical brain processing characteristics, which may contribute to their exceptional performance in sports.

AGCosPlace: A UAV Visual Positioning Algorithm Based on Transformer

To address the limitation and obtain the position of the drone even when the relative poses and intrinsics of the drone camera are unknown, a visual positioning algorithm based on image retrieval called AGCosPlace, which leverages the Transformer architecture to achieve improved performance, is proposed. Our approach involves subjecting the feature map of the backbone to an encoding operation that incorporates attention mechanisms, multi-layer perceptron coding, and a graph network module. This encoding operation allows for better aggregation of the context information present in the image. Subsequently, the aggregation module with dynamic adaptive pooling produces a descriptor with an appropriate dimensionality, which is then passed into the classifier to recognize the position. Considering the complexity associated with labeling visual positioning labels for UAV images, the visual positioning network is trained using the publicly available Google Street View SF-XL dataset. The performance of the trained network model on a custom UAV perspective test set is evaluated. The experimental results demonstrate that our proposed algorithm, which improves upon the ResNet backbone networks on the SF-XL test set, exhibits excellent performance on the UAV test set. The algorithm achieves notable improvements in the four evaluation metrics: R@1, R@5, R@10, and R@20. These results confirm that the trained visual positioning network can effectively be employed in UAV visual positioning tasks.

LPR: learning point-level temporal action localization through re-training

Point-level temporal action localization (PTAL) aims to locate action instances in untrimmed videos with only one timestamp annotation for each action instance. Existing methods adopt the localization-by-classification paradigm to locate action boundaries in the temporal class activation map (TCAM) by thresholding, also known as TCAM-based method. However, TCAM-based methods are limited by the gap between classification and localization tasks, since TCAM is generated by a classification network. To address this issue, we propose a re-training framework for the PTAL task, also known as LPR. This framework consists of two stages: pseudo-label generation and re-training. In the pseudo-label generation stage, we propose a feature embedding module based on a transformer encoder to capture global context features and optimize pseudo-labels’ quality by leveraging point-level annotations. In the re-training stage, LPR uses the above pseudo-labels as supervision to locate action instances with a temporal action localization network rather than generating TCAMs. Furthermore, to alleviate the effects of label noise in the pseudo-labels, we propose a joint learning classification module (JLCM) in the re-training stage. This module contains two classification sub-modules that simultaneously predict action categories and are guided by a jointly determined clean set for network training. The proposed framework achieves state-of-the-art localization performance on both the THUMOS’14 and BEOID datasets.

Concentrating to avoid falling: interaction between peripheral sensory and central attentional demands during a postural stability limit task in sedentary seniors.

Evidence suggests falls and postural instabilities among seniors are attributed to a decline in both the processing of afferent signals (e.g., proprioceptive, vestibular) and attentional resources. We investigated the interaction between the non-visual and attentional demands of postural control in sedentary seniors. Old and young adults performed a postural stability limit task involving a maximal voluntary leaning movement with and without vision as well as a cognitive-attentional subtraction task. These tasks were performed alone (single-task) or simultaneously (dual-task) to vary the sensory-attentional demands. The functional limits of stability were quantified as the maximum center of pressure excursion during voluntary leaning. Seniors showed significantly smaller limits of postural stability compared to young adults in all sensory-attentional conditions. However, surprisingly, both groups of subjects reduced their stability limits by a similar amount when vision was removed. Furthermore, they similarly decreased their anterior-posterior stability limits when concurrently performing the postural and the cognitive-attentional tasks with vision. The overall average cognitive performance of young adults was higher than seniors and was only slightly affected during dual-tasking. In contrast, older adults markedly degraded their cognitive performance from the single- to the dual-task situations, especially when vision was unavailable. Thus, their dual-task costs were higher than those of young adults and increased in the eyes-closed condition, when postural control relied more heavily on non-visual sensory signals. Our findings provide the first evidence that as posture approaches its stability limits, sedentary seniors allot increasingly large cognitive attentional resources to process critical sensory inputs.

MultiTab: A Novel Portable Device to Evaluate Multisensory Skills.

To infer spatial-temporal features of an external event we are guided by multisensory cues, with intensive research showing an enhancement in the perception when information coming from different sensory modalities are integrated. In this scenario, the motor system seems to also have an important role in boosting perception. With the present work, we introduce and validate a novel portable technology, named MultiTab, which is able to provide auditory and visual stimulation, as well as to measure the user's manual responses. Our preliminary results indicate that MultiTab reliably induces multisensory integration in a spatial localization task, shown by significantly reduced manual response times in the localization of audiovisual stimuli compared to unisensory stimuliClinical relevance- The current work presents a novel portable device that could contribute to the clinical evaluation of multisensory processing as well as spatial perception. In addition, by promoting and recording manual actions, MultiTab could be especially suitable for the design of rehabilitative protocols using multisensory motor training.

Effect of sit-stand workstation position and computer task on head and trunk postural sway and discomfort

Adjustable-height desks may provide musculoskeletal health benefits to offset the effects of prolonged sitting. One mechanism may be increased postural variability, here characterized by head and trunk postural sway. Linear acceleration of the head and trunk were measured while participants used computer workstations in seated and standing positions during keyboard and mouse tasks; secondary measures were discomfort and proprioception (head and neck repositioning error). Median accelerations of the head and trunk were 20–26% lower in mouse tasks compared to keyboard tasks (p < 0.01). There were no significant differences in sway parameters between seated and standing positions. Discomfort and proprioception were correlated; subjects who experienced increased neck discomfort after 1.5 h of computer work had almost twice the head and neck repositioning error. The results suggest that postural sway is more affected by different tasks (keyboard vs. mouse) than by different workstation configurations and that low proprioception acuity may relate to the development of discomfort.

Difference in the recruitment of intrinsic foot muscles in the elderly under static and dynamic postural conditions.

The effect of foot, especially intrinsic muscles, on postural control and its related mechanisms remain unclear due to the complex structure. Therefore, this study aims to investigate the activation of intrinsic foot muscles in the elderly under static and dynamic postural tasks. Twenty-one elderly participants were included to perform different postural tests (sensory organization test (SOT), motor control test (MCT), limit of stability test (LOS), and unilateral stance test) by a NeuroCom Balance Manager System. The participants were instructed to maintain postural stability under conditions with combined different sensory inputs (vision, vestibular, and proprioception) in SOT as well as conditions with translation disturbance in MCT, and to perform an active weight-shifting tasks in LOS. During these tasks, muscle activation were simultaneously acquired from intrinsic foot muscles (abductor halluces (AbH) and flexor digitorum brevis (FDB)) and ankle muscles (anterior tibialis, medial head of gastrocnemius, lateral head of gastrocnemius, and peroneus longus). The root-mean-square amplitude of these muscles in postural tasks was calculated and normalized with the EMG activity in unilateral stance task. The activation of intrinsic foot muscles significantly differed among different SOT tasks (p<0.001). Post-hoc tests showed that compared with that under normal condition 1 without sensory interference, EMGs increased significantly under sensory disturbance (conditions 2-6). By contrast, compared with that under the single-sensory disturbed conditions (conditions 2-4; 2 for disturbed vision, 3 for disturbed vestibular sensation, 4 for disturbed proprioception), activation was significantly greater under the dual-sensory disturbed postural tasks (conditions 5 and 6; 5 for disturbed vision and proprioception, 6 for disturbed vestibular sensation and proprioception). In MCT, EMGs of foot muscles increased significantly under different translation speeds (p<0.001). In LOS, moderate and significant correlations were found between muscle activations and postural stability parameters (AbH, r = 0. 355-0.636, p<0.05; FDB, r = 0.336-0.622, p<0.05). Intrinsic foot muscles play a complementary role to regulate postural stability when disturbances occur. In addition, the recruitment magnitude of intrinsic foot muscles is positively correlated with the limit of stability, indicating their contribution to increasing the limits of stability in the elderly.

Optimizing Appearance-Based Localization with Catadioptric Cameras: Small-Footprint Models for Real-Time Inference on Edge Devices.

This paper considers the task of appearance-based localization: visual place recognition from omnidirectional images obtained from catadioptric cameras. The focus is on designing an efficient neural network architecture that accurately and reliably recognizes indoor scenes on distorted images from a catadioptric camera, even in self-similar environments with few discernible features. As the target application is the global localization of a low-cost service mobile robot, the proposed solutions are optimized toward being small-footprint models that provide real-time inference on edge devices, such as Nvidia Jetson. We compare several design choices for the neural network-based architecture of the localization system and then demonstrate that the best results are achieved with embeddings (global descriptors) yielded by exploiting transfer learning and fine tuning on a limited number of catadioptric images. We test our solutions on two small-scale datasets collected using different catadioptric cameras in the same office building. Next, we compare the performance of our system to state-of-the-art visual place recognition systems on the publicly available COLD Freiburg and Saarbrücken datasets that contain images collected under different lighting conditions. Our system compares favourably to the competitors both in terms of the accuracy of place recognition and the inference time, providing a cost- and energy-efficient means of appearance-based localization for an indoor service robot.

Adaptation strategies and neurophysiological response in early-stage Parkinson's disease: BioVRSea approach.

There is accumulating evidence that many pathological conditions affecting human balance are consequence of postural control (PC) failure or overstimulation such as in motion sickness. Our research shows the potential of using the response to a complex postural control task to assess patients with early-stage Parkinson's Disease (PD). We developed a unique measurement model, where the PC task is triggered by a moving platform in a virtual reality environment while simultaneously recording EEG, EMG and CoP signals. This novel paradigm of assessment is called BioVRSea. We studied the interplay between biosignals and their differences in healthy subjects and with early-stage PD. Despite the limited number of subjects (29 healthy and nine PD) the results of our work show significant differences in several biosignals features, demonstrating that the combined output of posturography, muscle activation and cortical response is capable of distinguishing healthy from pathological. The differences measured following the end of the platform movement are remarkable, as the induced sway is different between the two groups and triggers statistically relevant cortical activities in α and θ bands. This is a first important step to develop a multi-metric signature able to quantify PC and distinguish healthy from pathological response.

Intention-reflected predictive display for operability improvement of time-delayed teleoperation system

Robotic teleoperation is highly valued for its ability to remotely execute tasks that demand sophisticated human decision-making or that are intended to be carried out by human operators from a distance. However, when using the internet as a communication framework for teleoperation, high latency, and fluctuations make accurate positioning and time-dependent tasks difficult. To mitigate the negative effects of time delay, this paper proposes a teleoperation system that uses cross reality (XR) as a predictive display of the outcome of operators’ intended actions and develops a time-delay aware shared control to fulfill the intention. The system targets a liquid pouring task, wherein a white ring that indicates the intended height of the liquid surface is overlayed onto the beaker in a delayed camera image to close the visual feedback loop on the leader side. Simultaneously, the shared control automatically completes the pouring action to track the intended liquid height. The performance of the proposed system is validated based on liquid pouring experiments performed by human subjects. When compared with direct control, the absolute error rate decreased significantly for a constant round-trip time delay of 0.8 s and 1.2 s, similarly for a time-varying delay of 0.4 s and 0.8 s. Moreover, when the time-varying delay was 0.8 s, operators achieved significantly higher accuracy while maintaining comparable operation time. These results indicate that our proposed system improves operability even in the presence of time-varying delays in communication networks.

Transcranial direct current stimulation suggests not improving postural control during adapted tandem position in people with Parkinson's disease: A pilot study

BackgroundBalance impairments in people with Parkinson's disease (PD) demonstrated mainly in challenging postural tasks, such as increased body oscillation may be attributed to the deficits in the brain structures functionality involved in postural control (e.g., motor cortex, midbrain, and brainstem). Although promising results, the effect of transcranial direct current stimulation (tDCS) on postural control in people with PD is unclear, especially in objective measures such as the center of pressure (CoP) parameters. Thus, we analyzed the effects of a single session of tDCS on the CoP parameters during the adapted tandem position in people with PD. MethodsNineteen people with PD participated in this crossover, randomized, and double-blind study. Anodal tDCS was applied over the primary motor cortex in two conditions of stimulation (2 mA/active and sham) on two different days for 20 min immediately before the postural control evaluation. Participants remained standing in an adapted tandem position for the postural control assessment for 30 s (three trials). CoP parameters were acquired by a force plate. ResultsNo significant differences were demonstrated between stimulation conditions (p-value range = 0.15–0.89). ConclusionsOur results suggested that a single session of tDCS with 2 mA does not improve the postural control of people with PD during adapted tandem.

Wi-Fi-Based Indoor Localization and Navigation: A Robot-Aided Hybrid Deep Learning Approach.

Indoor localization and navigation have become an increasingly important problem in both industry and academia with the widespread use of mobile smart devices and the development of network techniques. The Wi-Fi-based technology shows great potential for applications due to the ubiquitous Wi-Fi infrastructure in public indoor environments. Most existing approaches use trilateration or machine learning methods to predict locations from a set of annotated Wi-Fi observations. However, annotated data are not always readily available. In this paper, we propose a robot-aided data collection strategy to obtain the limited but high-quality labeled data and a large amount of unlabeled data. Furthermore, we design two deep learning models based on a variational autoencoder for the localization and navigation tasks, respectively. To make full use of the collected data, a hybrid learning approach is developed to train the models by combining supervised, unsupervised and semi-supervised learning strategies. Extensive experiments suggest that our approach enables the models to learn effective knowledge from unlabeled data with incremental improvements, and it can achieve promising localization and navigation performance in a complex indoor environment with obstacles.

Synthetic-real image domain adaptation for indoor camera pose regression using a 3D model

Deep learning-based camera pose regression approaches have achieved outstanding performance for visual indoor localisation. However, these approaches are limited by the availability of images with known camera poses, and they often require a comprehensive mapping of the indoor scenes, which is labour-intensive and often impractical. Recent studies have shown that synthetic images derived from simple 3D building models can be used to train deep learning models to perform cross-domain synthetic-to-real visual localisation. But the performance of such cross-domain localisation models is degraded due to the domain gap between the real and synthetic images. In this study, we propose a domain adaptation approach based on a Generative Adversarial Network (GAN) framework, which is trained on a set of unpaired synthetic and real images to generate real-looking synthetic images and synthetic-looking real images. These images are then used to train the current deep learning-based camera pose regression networks. We develop two domain adaptation strategies for the localisation task, namely, synthetic-to-real and real-to-synthetic image style transfer. The results show that the best model is obtained for the real-to-synthetic style transfer with a localisation error of 0.31 m and orientation error of 3.36°, which is approximately three times lower compared to the previously reported results.

Rambling and trembling trajectories in the analysis of postural sway prior to the self-paced and reaction time tasks

Objective: Previous studies suggested that center of pressure (COP) shifts occur before an expected perturbation in the form of early and anticipatory postural adjustments which operate in a short time scale. However, the effect of such perturbations on pre-existing postural set on a longer time scale remained uncovered. The purpose of this study was to investigate whether rambling and trembling components of the COP trajectories depend on postural task or phase of trial before a self-initiated perturbation. Materials and Methods: Twenty-four young healthy participants took part in the study. Subjects performed three postural tasks, namely, (i) quiet stance task: 60 seconds quiet stance, (ii) self-paced task: maximal vertical jump from quiet stance under the self-paced time condition, and (iii) reaction-time task: maximal vertical jump from quiet stance under the reaction-time condition. Postural sway features were examined in two phases, the first and last 20 seconds of the trials. Results: The features of rambling and trembling components of the COP trajectories were affected by postural task or phase of trial. The ellipse area of the COP and rambling trajectories were significantly different among postural tasks. The median frequency was significantly different between the phases of trials for the COP and rambling trajectories. Conclusion: This study indicated task-specific changes in postural sway features. Rambling and trembling trajectories, which would reflect two underlying human postural control mechanisms as maintaining the body's equilibrium with respect to a moving reference point and oscillating around the moving reference point respectively, were affected differently before a whole-body maximum-effort self-initiated perturbation.

Effects of neurofeedback on standing postural control task with combined imagined and executed movements.

Motor imagery (MI) is a method of imagining movement without actual movement, and its use in combination with motor execution (ME) enhances the effects of motor learning. Neurofeedback (NFB) is another method that promotes the effects of MI. This study aimed to investigate the effects of NFB on combined MI and ME (MIME) training in a standing postural control task. Sixteen participants were randomly divided into MIME and MIME + NFB groups and performed 10 trials of a postural control task on an unstable board, with nine trials of MI in between. Electroencephalogram was assessed during MI, and the MIME + NFB group received neurofeedback on the degree of MI via auditory stimulation. A postural control task using an unstable board was performed before and after the MIME task, during which postural instability was evaluated. Postural instability was reduced after the MIME task in both groups. In addition, the root mean square, which indicates the sway of the unstable board, was significantly reduced in the MIME + NFB group compared to that in the MIME group. Our results indicate that MIME training is effective for motor learning of standing postural control. Furthermore, when MI and ME are combined, the feedback on the degree of MI enhances the learning effect.

Multiattribute multitask transformer framework for vision‐based structural health monitoring

AbstractUsing deep learning (DL) to recognize building and infrastructure damage via images is becoming popular in vision‐based structural health monitoring (SHM). However, many previous studies solely work on the existence of damage in the images and directly treat the problem as a single‐attribute classification or separately focus on finding the location or area of the damage as a localization or segmentation problem. Abundant information in the images from multiple sources and intertask relationships are not fully exploited. In this study, the vision‐based SHM problem is first reformulated into a multiattribute multitask setting, where each image contains multiple labels to describe its characteristics. Subsequently, a general multiattribute multitask detection framework, namely ϕ‐NeXt, is proposed, which introduces 10 benchmark tasks including classification, localization, and segmentation tasks. Accordingly, a large‐scale data set containing 37,000 pairs of multilabeled images is established. To pursue better performance in all tasks, a novel hierarchical framework, namely multiattribute multitask transformer (MAMT2) is proposed, which integrates multitask transfer learning mechanisms and adopts a transformer‐based network as the backbone. Finally, for benchmarking purposes, extensive experiments are conducted on all tasks and the performance of the proposed MAMT2 is compared with several classical DL models. The results demonstrate the superiority of the MAMT2 in all tasks, which reveals a great potential for practical applications and future studies in both structural engineering and computer vision.

Interchangeable Role of Motor Cortex and Reafference for the Stable Execution of an Orofacial Action.

Animals interact with their environment through mechanically active, mobile sensors. The efficient use of these sensory organs implies the ability to track their position; otherwise, perceptual stability or prehension would be profoundly impeded. The nervous system may keep track of the position of a sensorimotor organ via two complementary feedback mechanisms-peripheral reafference (external, sensory feedback) and efference copy (internal feedback). Yet, the potential contributions of these mechanisms remain largely unexplored. By training male rats to place one of their vibrissae within a predetermined angular range without contact, a task that depends on knowledge of vibrissa position relative to their face, we found that peripheral reafference is not required. The presence of motor cortex is not required either, except in the absence of peripheral reafference to maintain motor stability. Finally, the red nucleus, which receives descending inputs from motor cortex and cerebellum and projects to facial motoneurons, is critically involved in the execution of the vibrissa positioning task. All told, our results point toward the existence of an internal model that requires either peripheral reafference or motor cortex to optimally drive voluntary motion.SIGNIFICANCE STATEMENT How does an animal know where a mechanically active, mobile sensor lies relative to its body? We address this basic question in sensorimotor integration using the motion of the vibrissae in rats. We show that rats can learn to reliably position their vibrissae in the absence of sensory feedback or in the absence of motor cortex. Yet, when both sensory feedback and motor cortex are absent, motor precision is degraded. This suggests the existence of an internal model able to operate in closed- and open-loop modes, requiring either motor cortex or sensory feedback to maintain motor stability.