Inverse Dynamics Research Articles

Assessment of stabilizing feedback control of walking: A tutorial

Walking without falling requires stabilization of the trajectory of the body center of mass relative to the base of support. Model studies suggest that this requires active, feedback control, i.e., the nervous system must process sensory information on the state of the body to generate descending motor commands to the muscles to stabilize walking, especially in the mediolateral direction. Stabilization of bipedal gait is challenging and can be impaired in older and diseased individuals. In this tutorial, we illustrate how gait analysis can be used to assess the stabilizing feedback control of gait. We present methods ranging from those that require limited input data (e.g. position data of markers placed on the feet and pelvis only) to those that require full-body kinematics and electromyography. Analyses range from simple kinematics analyses to inverse dynamics. These methods assess stabilizing feedback control of human walking at three levels: 1) the level of center of mass movement and horizontal ground reaction forces, 2) the level of center of mass movement and foot placement and 3) the level of center of mass movement and the joint moments or muscle activity. We show how these can be calculated and provide a GitHub repository (https://github.com/VU-HMS/Tutorial-stabilizing-walking) which contains open access Matlab and Python code to calculate these. Finally, we discuss what information on feedback control can be learned from each of these.

Numerical study of energy consumption in spacer-filled feed channels through 3D inverse modeling and computational fluid dynamics

This project employs advanced Computational Fluid Dynamics (CFD) simulations to investigate the complex flow field dynamics within the channel of membrane module featuring a commercially available feed spacer. Utilizing Micro-CT scanning, the intricate three-dimensional (3D) structure of the spacer is accurately captured, enhancing the precision of the simulation models. The project focuses on the effects of varying Reynolds number (Re) and Aspect Ratio (AR) on the energy consumption within the feed channel of membrane module, analyzing streamline patterns, vorticity, and flow separation angles in detail. Findings indicate that the total enstrophy identification method effectively pinpoints areas of high energy consumption, especially at lower Re levels where flow separation significantly contributes to hydraulic losses. As Re increases, the separation vortex shifts rearward, resulting in a marked decrease in the energy consumption coefficient. Furthermore, the flow separation angle consistently declines with increasing Re, stabilizing at higher levels. Significantly, a reduction in AR from 1.015 to 0.725 more than doubles the decrease in flow separation angle, highlighting substantial potential energy savings through spacer design optimization. These insights are crucial for improving the design and efficiency of spiral wound membrane modules and for refining feed spacer configurations to enhance performance.

Adaptive neural network control for nonholonomic systems with time-varying asymmetric constraints and iISS inverse dynamics

Adaptive neural network control for nonholonomic systems with time-varying asymmetric constraints and iISS inverse dynamics

Bridging the sim2real gap. Investigating deviations between experimental motion measurements and musculoskeletal simulation results-a systematic review.

Musculoskeletal simulations can be used to estimate biomechanical variables like muscle forces and joint torques from non-invasive experimental data using inverse and forward methods. Inverse kinematics followed by inverse dynamics (ID) uses body motion and external force measurements to compute joint movements and the corresponding joint loads, respectively. ID leads to residual forces and torques (residuals) that are not physically realistic, because of measurement noise and modeling assumptions. Forward dynamic simulations (FD) are found by tracking experimental data. They do not generate residuals but will move away from experimental data to achieve this. Therefore, there is a gap between reality (the experimental measurements) and simulations in both approaches, the sim2real gap. To answer (patho-) physiological research questions, simulation results have to be accurate and reliable; the sim2real gap needs to be handled. Therefore, we reviewed methods to handle the sim2real gap in such musculoskeletal simulations. The review identifies, classifies and analyses existing methods that bridge the sim2real gap, including their strengths and limitations. Using a systematic approach, we conducted an electronic search in the databases Scopus, PubMed and Web of Science. We selected and included 85 relevant papers that were sorted into eight different solution clusters based on three aspects: how the sim2real gap is handled, the mathematical method used, and the parameters/variables of the simulations which were adjusted. Each cluster has a distinctive way of handling the sim2real gap with accompanying strengths and limitations. Ultimately, the method choice largely depends on various factors: available model, input parameters/variables, investigated movement and of course the underlying research aim. Researchers should be aware that the sim2real gap remains for both ID and FD approaches. However, we conclude that multimodal approaches tracking kinematic and dynamic measurements may be one possible solution to handle the sim2real gap as methods tracking multimodal measurements (some combination of sensor position/orientation or EMG measurements), consistently lead to better tracking performances. Initial analyses show that motion analysis performance can be enhanced by using multimodal measurements as different sensor technologies can compensate each other's weaknesses.

Finite-time stability of a class of nonlinear cascaded systems with time-varying output constraints

The finite-time stabilization problem of a class of cascaded systems with time-varying output constraints is investigated in this paper. The system consists of input-to-state stable inverse dynamics and a nonholonomic system. The technology of changing the supply rate eliminates dynamic uncertainty. A new tan-type barrier Lyapunov function is utilized to handle time-varying output constraints. Under the unified framework of constrained and unconstrained systems, combined with the addition of power integrator technology and switch control strategy, a state feedback control strategy is proposed to ensure that the state of the closed-loop system converges to zero in a finite-time and always satisfies time-varying output constraints. Finally, simulation results verify the proposed control scheme’s effectiveness.

DynaSTI: Dynamics modeling with sequential temporal information for reinforcement learning in Atari

Deep reinforcement learning (DRL) has shown remarkable capabilities in solving sequential decision-making problems. However, DRL requires extensive interactions with image-based environments. Existing methods have combined self-supervised learning or data augmentation to improve sample efficiency. While understanding the temporal information dynamics of the environment is important for effective learning, many methods do not consider these factors. To address the sample efficiency problem, we propose dynamics modeling with sequential temporal information (DynaSTI) that incorporates environmental dynamics and leverages the correlation among trajectories to improve sample efficiency. DynaSTI uses an effective learning strategy for state representation as an auxiliary task, using gated recurrent units to capture temporal information. It also integrates forward and inverse dynamics modeling in a hierarchical configuration, enhancing the learning of environmental dynamics compared to using each model separately. The hierarchical structure of DynaSTI enhances the stability of inverse dynamics modeling during training by using inputs derived from forward dynamics modeling, which focuses on feature extraction related to controllable state. This approach effectively filters out noisy information. Consequently, using denoised inputs from forward dynamics modeling results in improved stability when training inverse dynamics modeling, rather than using inputs directly from the encoder. We demonstrate the effectiveness of DynaSTI through experiments on the Atari game benchmark, limiting the environment interactions to 100k steps. Our extensive experiments confirm that DynaSTI significantly improves the sample efficiency of DRL, outperforming comparison methods in terms of statistically reliable metrics and nearing human-level performance.

Research and Experiment on Cruise Control of a Self-Propelled Electric Sprayer Chassis

In order to address the issues of poor stability in vehicle speed and deteriorated spraying quality caused by changes in road slope and the decrease in overall mass due to liquid spraying, this study focuses on analyzing the structure and longitudinal dynamic characteristics of a 4WID high ground clearance self-propelled electric sprayer. By utilizing MATLAB/Simulink software, three subsystems, namely, the inverse longitudinal dynamics model, torque distribution model, and motor model, are established. The model takes into account the effects of longitudinal driving resistance, slope, and vehicle roll angle on the distribution of loads among the four wheels during slope driving. A seven-degrees-of-freedom dynamic model is developed. A hierarchical control structure is designed, incorporating an upper-level PID controller and a lower-level fuzzy PID controller, to control the overall system. The control algorithms are tailored to the specific characteristics of the sprayer’s operation, and simulation experiments are conducted under the corresponding operating conditions. Building upon this, a sensor-equipped experimental platform is set up in the self-propelled sprayer manufactured by the team in the preliminary stage. Real vehicle tests are conducted in two scenarios: transition transportation and field operations, with the evaluation of the overall vehicle speed serving as the performance metric to validate the correctness of the model and the control theory.

Influence of restricted visual input on lower limb joint works of female children during sit-to-stand

BackgroundThe ankle-knee-hip joint systems have structures that can produce mechanical work through elastic, viscoelastic mechanisms or muscle activity. This study aimed to compute sit-to-stand (STS) joint works in lower limbs between blind and sighted children to find the relationship between visual memory and STS joint work variables. MethodsThis study included fifteen female children with congenitally blind (CB) and 30 healthy girls without visual impairments. The children with no visual impairments were randomly divided into two condition groups with 15 each, the eyes open (EO) and the eyes closed (EC). Inverse dynamics calculated joint works by integrating multiple the moment and angular velocity (F1) and force and velocity (F2). They were normalized to body mass and body height. ResultsGenerally, the sensitivity of F1 (on both sides in the sagittal and frontal planes) was more than F2 (on the non-dominant side in the mediolateral and vertical axes). In the ML axis, the EC group had insufficient maximal non-dominant hip work relative to the EO group (p = 0.002). In addition, the CB group suffered from low hip efficiency (p = 0.003) and high knee (p < 0.001) mechanical work. ConclusionsNumerous differences between CB and EC groups (on knee and hip works) showed that the time of visual input deprivation could change the type of human body's strategies to reach the consolidation process and keep adequate balance during STS. Therefore, rehabilitation programs should be aimed at addressing the impairments in the management of restricted visual input during STS performance.

Formation of Flight Trajectories of an Unmanned Glider Based on the Methodology of Inverse Dynamics Problems

Formation of Flight Trajectories of an Unmanned Glider Based on the Methodology of Inverse Dynamics Problems

A Numerical Approximation Approach for Deriving Computational Efficient Inverse Dynamics of 6-DOF Parallel Robots Based on Principle of Virtual Work

A Numerical Approximation Approach for Deriving Computational Efficient Inverse Dynamics of 6-DOF Parallel Robots Based on Principle of Virtual Work

Integrating long short-term memory for optimal control of 6-DOF welding robot arm

In the realm of mechatronics, robots stand out as emblematic manifestations of societal progress. The synergy between the evolution of robotic technologies and artificial intelligence (AI) has been pivotal in refining performance and elevating the automation capabilities of robotic arms. This study delves into the integration of the long short-term memory (LSTM) AI model in the control of the inverse dynamics of a six-degree-of-freedom welding robot arm, employing a velocity-based motion control methodology. Noteworthy for its innovative operational approach, this method is widely employed across diverse models of industrial robotic arms. The research findings underscore the superior optimization results achieved by the LSTM model during the accumulation of control signals, surpassing previous studies in the same domain. Anticipated as a catalyst for substantial improvements in the efficiency of welding robot operations, this model heralds a promising avenue for future advancements in optimization techniques.

Biomechanical improvements in gait following medial pivot knee implant surgery

BackgroundTotal knee replacements are used to improve function and reduce pain in patients with advanced osteoarthritis. The medially stabilising implant is designed to mimic a healthy knee. This study aims to provide a comprehensive analysis of the kinematics and kinetics of a medially stabilising knee implant, comparing it to a healthy control group, as well as to its pre-operative state and the contralateral limb. MethodsSixteen total knee replacement patients and ten healthy participants were recruited. Patients underwent testing 4–6 weeks before surgery and repeated the same tests 12 months after surgery. Healthy participants completed the same tests at a single time point. All participants completed three walking trials: kinematics was captured with eight cameras; kinetics with in-ground force plates. Subject-specific musculoskeletal models were developed in OpenSim. Inverse kinematics and inverse dynamics were used to determine gait parameters. Joint angles and joint moments were evaluated using Statistical Parametric Mapping. Patient-reported outcome measures were also collected at both time points. FindingsSpatiotemporal results indicate significant differences in velocity and step length between pre-operative patients and control participants. Differences are observed in the adduction angles between the contralateral and ipsilateral limbs pre-operatively. Postoperatively, there was an increase in the 1st peak flexion moment, reduced adduction moment and reduced internal rotation moment. In PROMs, patients all report improvements in pain levels and high satisfaction levels following surgery. InterpretationsFollowing medial stabilising total knee arthroplasty, patients displayed improved clinical parameters and joint moments reflecting a shift towards more normal, healthy gait.

Calibrationless monocular vision musculoskeletal simulation during gait

With computer vision technology and prediction of ground reaction forces (GRF), a previous study performed markerless motion capture and musculoskeletal simulation with two smartphones (OpenCap). A recent approach can reconstruct 3D human motion from a single video without calibration and it may further simplify the motion capture process. However it has not been combined with musculoskeletal simulation and the validity is unclear. Therefore, the purpose of this study was to determine the validity of the musculoskeletal simulation using a monocular vision approach. An open-source dataset that contains motion capture and video data during gait from 10 healthy participants was used. Human motion reconstruction with the skinned human (SMPL) model was performed on each video. Virtual marker data was generated by extracting the position data from the SMPL skin vertices. Inverse kinematics, GRF prediction (only for monocular vision approach), inverse dynamics and static optimization were performed using a musculoskeletal model for experimental motion capture data and the generated virtual markers from videos. Mean absolute errors (MAE) between motion capture based and monocular vision based simulation outcomes were calculated. The MAE were 8.4° for joint angles, 5.0 % bodyweight for GRF, 1.1 % bodyweight*height for joint moments and 0.11 for estimated muscle activations from 16 muscles. The entire MAE was larger but some were comparable to OpenCap. Using the monocular vision approach, motion capture and musculoskeletal simulation can be done with no preparations and is beneficial for clinicians to quantify the daily gait assessment.

#1994 Analysis of dynamics of parameters of safety and effectiveness of rivaroxaban in patients with stage IV chronic kidney disease and atrial fibrillation

Abstract Background and Aims Research in the field of comorbidity of chronic kidney disease (CKD) and atrial fibrillation (AF) is becoming increasingly relevant, especially given the continuously increasing prevalence of this arrhythmia and its contribution to the risk of adverse events, the unity of factors in the development and progression of these diseases. This comorbidity is accompanied by an increased risk of thromboembolic complications (TEC) and atherothrombotic events, as well as the incidence of bleeding during anticoagulant therapy (ACT), which undoubtedly underlies the higher rates of cardiovascular and overall mortality in this cohort of patients. In this regard, we set the following goal: analysis of the dynamics and assessment of the safety parameters and effectiveness of the use of rivaroxaban in patients with a glomerular filtration rate (GFR) of 15–29 ml/min/1.73 m2 with non-valvular atrial fibrillation. Method The medical records of 12,026 patients were analyzed; non-valvular AF was diagnosed in 3,715 cases. 502 (15%) patients had stage IV CKD or a sustained decrease in GFR to 15–29 l/min/1.73 m2 during hospitalization. Of these, 109 (3%) patients met the inclusion criteria and were randomized 2:1 to rivaroxaban 15 mg (n = 73) or warfarin (n = 36). The average follow-up period was 18 months. Of the 109 included patients, 2 refused to participate in the study, contact with 9 was lost, 8 patients started taking another direct oral anticoagulant, CKD progressed to stage V in 4 patients, and antiplatelet agents were added in 6 cases. Results To assess the dynamics of changes in creatinine levels and GFR, mixed linear models were constructed, where the individual dynamics of each patient's indicators were selected as random effects, and time and drug intake were selected as fixed effects. A graphical interpretation of the dynamics and differences is presented in Figs 1 and 2. There was a significant improvement in the dynamics of creatinine and GFR in the rivaroxaban group compared to the warfarin group: while patients receiving rivaroxaban had an increase in GFR and a decrease in creatinine levels, patients taking warfarin, on the contrary, showed inverse dynamics in renal function indicators. It should be noted that during the study, a transition from stage IV CKD to stage III was recorded in 43% of patients in the rivaroxaban group and in 34% in the warfarin group (p = 0.26). There were no significant differences in the incidence of any acute cerebrovascular accident, myocardial infarction and all-cause mortality (warfarin—1.4%; 6.9%; 6.8% versus rivaroxaban—5.6%; 2.85; 8.3% (p = 0.25; 0.66; 0.78). Patients taking warfarin were significantly more likely to develop minor bleeding compared to rivaroxaban (36.9% versus 61.1% (p = 0. 01)) and all clinically significant bleeding (10.9% vs. 27.7% (p = 0.03)) according to the ISTH scale. In the warfarin group, TTR&amp;gt;70% was achieved in 94% of patients. All-cause readmission rates were 43% in the rivaroxaban group and 48% in the warfarin group (p = 0.57). Among these patients, 36.9% and 40.7% (respectively) were hospitalized for emergency reasons (p = 0.96). Conclusion In patients with atrial fibrillation and a low glomerular filtration rate (eGFR &amp;lt;30 ml/min/1.73 m2), it is reasonable to consider the use of rivaroxaban based on the current study showing a slower rate of decline in renal function, as well as favorable efficacy and safety profiles of rivaroxaban compared with warfarin in this group of patients.

Adaptive output feedback stabilisation control for cascaded systems with output constraints and disturbance

The system under consideration is a cascaded system composed of integral input-to-state stable inverse dynamics and a non-holonomic system. An adaptive output feedback stabilisation control is proposed for the system with disturbance and output constraints. The dynamic uncertainty is eliminated using the technique that changes the supply rate. An extended state observer is proposed by defining the non-vanishing disturbance as a new state variable of the system. The asymmetric time-varying output constraint is solved by utilising the tan-type barrier Lyapunov function. The result shows that the adaptive output feedback controller guarantees the stability of the closed-loop system without violating the asymmetric time-varying output constraints under the backstepping technique. Simulation results are given to verify the effectiveness of the control method.

Inverting a model of neuromuscular control to estimate descending activation patterns that generate fast-reaching movements.

Reaching movements generally show smooth kinematic profiles that are invariant across varying movement speeds even as interaction torques and muscle properties vary nonlinearly with speed. How the brain brings about these invariant profiles is an open question. We developed an analytical inverse dynamics method to estimate descending activation patterns directly from observed joint angle trajectories based on a simple model of the stretch reflex, and of muscle and biomechanical dynamics. We estimated descending activation patterns for experimental data from eight different planar two-joint movements performed at two movement times (fast: 400 ms; slow: 800 ms). The temporal structure of descending activation differed qualitatively across speeds, consistent with the idea that the nervous system uses an internal model to generate anticipatory torques during fast movement. This temporal structure also depended on the cocontraction level of antagonistic muscle groups. Comparing estimated muscle activation and descending activation revealed the contribution of the stretch reflex to movement generation that was found to set in after about 20% of movement time.NEW & NOTEWORTHY By estimating descending activation patterns directly from observed movement kinematics based on a model of the dynamics of the stretch reflex, of muscle force generation, and of the biomechanics of the limb, we observed how brain signals must be temporally structured to enable fast movement.

Prosthetic Limb Attachment via Electromagnetic Attraction Through a Closed Skin Envelope.

Current socket-based methods of prosthetic limb attachment are responsible for many of the dominant problems reported by persons with amputation. In this work, we introduce a new paradigm for attachment via electromagnetic attraction between a bone-anchored ferromagnetic implant and an external electromagnet. Our objective was to develop a design framework for electromagnetic attachment, and to evaluate this framework in the context of transfemoral amputation. We first used inverse dynamics to calculate the forces required to suspend a knee-ankle-foot prosthesis during gait. We then conducted cadaveric dissections to inform implant geometry and design a surgical methodology for covering the implant. We also developed an in silico framework to investigate how electromagnet design affects system performance. Simulations were validated against benchtop testing of a custom-built electromagnet. The physical electromagnet matched simulations, with a root-mean-square percentage error of 4.2% between measured and predicted forces. Using this electromagnet, we estimate that suspension of a prosthesis during gait would require 33 W of average power. After 200 and 1000 steps of simulated walking, the temperature at the skin would increase 2.3 °C and 15.4 °C relative to ambient, respectively. Our design framework produced an implant and electromagnet that could feasibly suspend a knee-ankle-foot prosthesis during short walking bouts. Future work will focus on optimization of this system to reduce heating during longer bouts. This work demonstrates the initial feasibility of an electromagnetic prosthetic attachment paradigm that has the potential to increase comfort and improve residual limb health for persons with amputation.

Model-Based Reinforcement Learning With Isolated Imaginations.

World models learn the consequences of actions in vision-based interactive systems. However, in practical scenarios like autonomous driving, noncontrollable dynamics that are independent or sparsely dependent on action signals often exist, making it challenging to learn effective world models. To address this issue, we propose Iso-Dream++, a model-based reinforcement learning approach that has two main contributions. First, we optimize the inverse dynamics to encourage the world model to isolate controllable state transitions from the mixed spatiotemporal variations of the environment. Second, we perform policy optimization based on the decoupled latent imaginations, where we roll out noncontrollable states into the future and adaptively associate them with the current controllable state. This enables long-horizon visuomotor control tasks to benefit from isolating mixed dynamics sources in the wild, such as self-driving cars that can anticipate the movement of other vehicles, thereby avoiding potential risks. On top of our previous work (Pan et al. 2022), we further consider the sparse dependencies between controllable and noncontrollable states, address the training collapse problem of state decoupling, and validate our approach in transfer learning setups. Our empirical study demonstrates that Iso-Dream++ outperforms existing reinforcement learning models significantly on CARLA and DeepMind Control.

Impact of model geometry and joint center locations on inverse kinematic/dynamic predictions: A comparative study of sexually dimorphic models

This work illustrates the sensitivity of demographically characteristic body segment inertial properties and subject-specific customization on model performance. One characteristic demographic, gender, and one subject-specific characteristic, hip joint center location, were represented with musculoskeletal modeling to evaluate how design decisions may alter model outputs. Generic sexually dimorphic musculoskeletal models were developed from the commonly used Rajagopal model using male and female data adapted by Dumas et al. Hip joint centers of these models were adjusted based on functional joint center testing. The kinematics and dynamics of 40 gait cycles from four subjects are predicted using these models. Two-way analysis of variance (ANOVA) was performed on the continuous time series data using statistical parametric mapping (SPM) to assess changes in kinematics/dynamics due to either choice in model (Rajagopal vs Dumas) or whether joint center adjustment was performed. The SPM based two-way ANOVA of the inverse dynamics found that differences in the Rajagopal and Dumas models resulted in significant differences in sagittal plane moments during swing (0.115 ± 0.032 Nm/kg difference in mean hip flexion moment during initial swing and a 0.077 ± 0.041 Nm/kg difference in mean hip extension moment during terminal swing), and differences between the models with and without hip joint center adjustment resulted in significant differences in hip flexion and abduction moments during stance (0.217 ± 0.055 Nm/kg increased mean hip abductive moment). By comparing the outputs of these differently constructed models with each other, the study finds that dynamic predictions of stance are sensitive to positioning of joint centers, and dynamic predictions of swing are more sensitive to segment mass/inertial properties.

Sources of wintertime PM2.5 at a major city in an Alpine Valley: The role of atmospheric dispersion and inversion dynamics

Urban areas in mountain environments are generally located on valley floors surrounded by slopes, where mountain orography drives peculiar meteorology and atmospheric circulation. Also, persistent inversion dynamics may occur strongly affecting air pollution. This study characterizes the PM2.5 pollution in a major city located in an Alpine valley (Belluno, Northeastern Italy) during the cold season (Autumn-Winter). Major particulate species (elemental and organic carbon, major inorganic ions) and minor/trace elements conventionally used as tracers for source apportionment were analyzed, including oxalate and specific PM2.5-bound tracers for biomass burning (K+, levoglucosan, mannosan, galactosan) and for primary biogenic organic aerosol (arabitol, mannitol, glucose). PM2.5 sources are identified through positive matrix factorization and a series of post-processing tools. Results indicate that biomass burning, mostly emitted by residential wood combustion for domestic heating, is the major source (52% PM2.5 mass concentration), followed by secondary aerosol (21%), biogenic aerosol (20%), traffic (4%), and dust resuspension (3%). The source contributions are assessed by accounting for the local meteorology. Insights into the dispersion or buildup of PM2.5 sources were then investigated by dispersion normalization. In addition, the possible effects of persistent thermal inversion events were evaluated by assessing the inversion strength from temperature profiles measured at multiple weather stations at different elevations with respect to the source contributions. Beyond the identification of those emission sources requiring further mitigation actions, this study also analyses the potential effects of local meteorology on PM2.5 pollution. The methodologies applied in this study can be easily adopted to other mountain environments for successful management of air quality.