Low-level Control Research Articles

Vision-state Fusion: Improving Deep Neural Networks for Autonomous Robotics

Vision-based deep learning perception fulfills a paramount role in robotics, facilitating solutions to many challenging scenarios, such as acrobatic maneuvers of autonomous unmanned aerial vehicles (UAVs) and robot-assisted high-precision surgery. Control-oriented end-to-end perception approaches, which directly output control variables for the robot, commonly take advantage of the robot’s state estimation as an auxiliary input. When intermediate outputs are estimated and fed to a lower-level controller, i.e., mediated approaches, the robot’s state is commonly used as an input only for egocentric tasks, which estimate physical properties of the robot itself. In this work, we propose to apply a similar approach for the first time – to the best of our knowledge – to non-egocentric mediated tasks, where the estimated outputs refer to an external subject. We prove how our general methodology improves the regression performance of deep convolutional neural networks (CNNs) on a broad class of non-egocentric 3D pose estimation problems, with minimal computational cost. By analyzing three highly-different use cases, spanning from grasping with a robotic arm to following a human subject with a pocket-sized UAV, our results consistently improve the R2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^{2}$$\\end{document} regression metric, up to +0.51, compared to their stateless baselines. Finally, we validate the in-field performance of a closed-loop autonomous cm-scale UAV on the human pose estimation task. Our results show a significant reduction, i.e., 24% on average, on the mean absolute error of our stateful CNN, compared to a State-of-the-Art stateless counterpart.

The Design of a Speech Recognition-Based Finger Training Rehabilitation Device: Mechanical and System Aspects

This research introduces a novel speech recognition-based control system for finger rehabilitation aimed at addressing the inefficiencies and prolonged recovery periods commonly associated with traditional stroke rehabilitation methods. The system enables stroke patients to engage in finger exercises while simultaneously providing real-time feedback on the finger's angle, speed, and position. This feedback is invaluable for rehabilitation physicians, offering critical data for assessing and refining rehabilitation strategies. The system architecture is composed of three main components: the hardware circuitry, a lower-level computer control system, and a voice recognition-based human-computer interaction system, all integrated with a finger movement perception system. Employing the Hidden Markov Model (HMM) for pattern recognition in the voice interaction component, the system has undergone simulation testing to verify its effectiveness. The findings confirm that the speech recognition-based rehabilitation training system meets all design expectations, providing a safe, reliable, and promising approach for enhancing finger rehabilitation practices.

Optimal Observer-Based Power Imbalance Allocation for Frequency Regulation in Shipboard Microgrids

This paper proposes a two-level control strategy based on a super-twisting sliding-mode algorithm (STA) to optimally allocate power imbalances in shipboard microgrids (SMGs) while achieving frequency regulation. The strategy employs an STA observer to estimate the unknown power load demand imbalances in finite time. This estimate is then passed to an online high-level optimal control framework to periodically determine the optimal sequence of power reference values for each energy storage device (ESS), minimising the operational cost of the SMG. The online optimised power reference values are interpolated and passed to the low-level STA control strategy to control the output power of each ESS. The efficacy of the proposed methods is demonstrated through numerical simulations conducted on a prototypical model of an SMG equipped with two ESSs, namely batteries and fuel cells with associated hydrogen storage.

Energy management control strategies addressing the rSOC degradation phenomena in a polygeneration microgrid

Economically competitive long-term energy provision and performance stability are becoming increasingly strategic for the upcoming energy market. Despite hydrogen-based technologies representing a viable way for the aforementioned targets, such systems are affected by performance reduction over time. This paper is aimed at presenting energy management mitigation policies for a reversible solid oxide-based (rSOC) polygeneration microgrid. Particularly, interest has been paid towards the rSOC degradation phenomena and its mitigation by applying strategies allowing meeting target performance and satisfying economics. In a multilevel framework, the rSOC degradation mitigation has been pursued proposing two central-level strategies, namely Grid support (GS) and Limited Grid support (LGS), which act between the power split targets coming from the supervisory level and working set points actuated by the low-level controllers. The former aims at compensating for the energy gap caused by degradation relying on external contributions: power grid network and boilers. Alike, the goal of the latter is to keep the rSOC turned on as much as possible, therefore, complying with the energy management goals given by supervisory level. The state of charge of the hydrogen storage tank has been chosen as a metric to best show the results obtained implementing the mitigation strategies, as it directly displays the rSOC active management and boundaries applied on the storage systems. Considering a 3 years scenario, the LGS strategy in comparison to the GS one has resulted to be more effective concerning: costs, microgrid self-sufficiency, reliability and overall robustness. Quantitatively, for the LGS strategy, considering degradation degrees in the range 0.25–1 %/kh, relative additional integration costs with respect to the case unaffected by degradation are seen to be in the range 30–105 %.

Perceived parental bonding and cortisol awakening response in people with eating disorders

IntroductionEarly life experiences may have an impact on hypothalamic–pituitary–adrenal (HPA) axis functioning in eating disorders (EDs). Parental bonding is defined as the parental contribution of care and control to parent–child relationships. We evaluated whether perceived care and protection of parental bonding in childhood and adolescence were associated with HPA axis functioning in adult patients with EDs. The activity of the HPA axis was assessed by measuring the salivary cortisol awakening response (CAR).ObjectivesWe evaluated whether parental care and control in childhood and adolescence were associated with HPA axis functioning in adults with EDs. On the basis of literature data on healthy participants, we hypothesized that people with high levels of parental care would show a reduced CAR compared to people with low levels of parental care.MethodsWe admitted patients according to the following inclusion criteria: (a) female sex, (b) age > 18 years, (c) current diagnosis of AN or BN according to DSM-5 criteria, (d) absence of severe physical disorders, (e) no history of endocrine disorders, psychoactive substance use, schizophrenia or other psychoses, bipolar disorders or head trauma. Participants completed the Italian version of the Parental Bonding Instrument (PBI). To measure the CAR, participants were instructed to collect saliva samples at awakening and 15, 30, and 60 min after awakening.Results64 women with EDs participated in the study: 37 with AN and 27 with BN. 28 participants reported low levels of both maternal and paternal care while 12 participants reported high levels of care from both parents; 31 participants reported high levels of both maternal and paternal control, while 12 participants reported low levels of control from both parents. When maternal care was entered as between factor in the analysis, the repeated measures 3-way ANOVA showed a significant time effect (F3, 180 = 2.81, p < 0.05) and a significant maternal care X time interaction (F3, 180 = 2.99, p < 0.05), while, when paternal care was entered as between factor, the repeated measures 3-way ANOVA did not show significant effects for time and no significant paternal care X time interaction.ConclusionsOur results show an association of perceived maternal care with the time pattern of CAR in female patients with ED, while perceived parental control was not associated with any CAR feature in EDs. Maternal control, paternal care and paternal control were not associated with any difference in the CAR.Disclosure of InterestNone Declared

An inertial control method for large-scale wind farm based on hierarchical distributed hybrid deep-reinforcement learning

Wind farm (WF) is gradually required to provide inertial support during frequency events nowadays. Existing WF inertial control methods have exhibited limitations in parameter estimation and optimal control parameter tuning under variable operating conditions. Moreover, traditional model-based methods suffer from computational efficiency problems due to their reliance on repeated iterations or derivative computations in optimization process. Thus, this paper proposes an inertial control method for large-scale WF based on hierarchical distributed hybrid deep-reinforcement learning (HDH-DRL). Firstly, the control objectives for the inertial control are defined, and are decomposed on the basis of wind turbines (WTs) division. Next, this paper presents a HDH-DRL based inertial control framework consisting of two levels. The upper-level control is achieved by the hybrid multi-agent DRL (HMA-DRL) algorithm with hybrid action exploration mechanism and multi-agent coordination. The consensus based lower-level control aims to achieve the consensus convergence of the control through the locally distributed interaction among WTs. Finally, the inertial control processes of the model are exhibited; the influences of DRL algorithms on the control are discussed; the computational speed and solution accuracy of the proposed method are compared with the model-based methods; the multi-scenario applicability and performance in local communication failures of the model are analyzed. The results demonstrate that the proposed method significantly improves the active power response and frequency stability, achieving rapid consensus convergence with a power deviation below 0.3%, surpassing traditional control methods by at least 0.2%; also, it demonstrates superior performance across various scenarios, including transient and steady-state conditions, with frequency enhancements up to 0.30% and exceptional stability under different wind conditions and communication disruptions.

基于YOLOv7-tiny模型的马铃薯播种机漏重播检测系统的设计与试验

In response to the issues of missed and repeated planting during the operation of the chain-spoon type potato planter in China, as well as the low recognition rate for missed planting and the difficulty in identifying repeated planting using existing detection methods, an innovative Potato Planter Missed and Repeated Planting Detection System has been designed. This system is built with a PLC as the lower-level controller and an industrial computer as the core, incorporating the YOLO object detection algorithm for detecting missed and repeated plantings during the operation of the potato planter. Using the YOLOv7-tiny object detection network model as the core, and combining model training with hardware integration, the system performs real-time detection of the potato seed situation within the seed spoon during the operation of the potato planter. It can quickly distinguish between normal planting, missed planting, and repeated planting scenarios. By incorporating the working principles of the planter, the system designs a positioning logic to identify the actual coordinates of missed and repeated planting locations when a lack or excess of planting is detected. This is achieved through the positioning module, enhancing the system's capability to accurately obtain coordinate information for actual missed and repeated planting positions. The system was deployed and tested on a 2CM-2C potato planter. The results indicate that the detection accuracy for missed and repeated plantings reached 96.07% and 93.98%, respectively. Compared to traditional sensor detection methods, the system improved the accuracy of missed planting detection by 5.29%. Additionally, it successfully implemented the functionality of detecting repeated plantings, achieving accurate monitoring of quality-related information during the operation of the potato planter.

Spatiotemporal Distribution Characteristics and Inventory Analysis of Near-Road Traffic Pollution in Urban Areas

Vehicle emissions belong to the category of near-surface sources, occur close to human activity areas, and pose a greater threat to human health than other anthropogenic pollution sources. Furthermore, the study of the spatiotemporal characteristics of near-road traffic pollution is of great significance to urban and regional ambient air quality management, and is also an important basis for vehicle emission inventories, as well as the assessment of ambient air impact. Most previous studies have analyzed the spatiotemporal characteristics of hydrocarbons (HCs), carbon monoxide (CO), nitrogen oxides (NOx), and carbon dioxide (CO2) in urban vehicle emissions over a certain time, without considering the synergistic effect of mobile source particulate matter, NOx, and volatile organic compounds (VOCs). In this study, we analyze the composition of vehicles with different emission standards from road mobile sources in Anqing City, China. National category III and IV vehicles are the main contribution sources of various pollutants, accounting for more than 60% of emissions. Although national category I and II vehicles accounted for less than 1% of the total number of vehicles, their contribution to emissions cannot be ignored, especially for CO and HCs, the contribution of which from such vehicles can reach about 7%. This is mainly due to the low level of pollution control arising from the larger emission factor and greater age of these vehicles. Furthermore, eliminating old cars and increasing the proportion of national category VI vehicles can effectively reduce vehicle pollutant emissions. In terms of the spatiotemporal distribution characteristics, highways around urban areas are also the main sources of heavy vehicles, and the emission intensity of these pollutants is also higher on national roads and highways surrounding urban areas. In addition, the presence of m/p-xylene and toluene solvent-using species is detected, which indicates that petrol vehicle emissions, LPG and petrol volatilization, and solvent-using sources contribute significantly to ozone formation in the ozone pollution process. Comparing weekdays and non-weekdays, the PM2.5 peaks on non-weekdays are significantly higher than those on weekdays, and there is no “weekend effect”, which indicates that traffic emissions have little influence on PM2.5 emissions, and may be related to energy use and industrial pollution. Overall, this study strengthens the understanding of the relationship between emissions, traffic volumes, and vehicle types on spatial and temporal scales, and emphasizes the need for further investigation and comprehensive measures to mitigate pollution from these emissions.

Automatic alignment of underwater snake robots operating in wakes of bluff bodies

This paper presents the development of controllers to automatically align an underwater snake robot (USR) with a wake that forms behind a bluff body, while swimming at a desired distance to the object. The low-level controllers stabilize the joint motion of the USR to a swimming gait while achieving a desired orientation and tangential velocity. The high-level controllers are designed to select references for the orientation and tangential velocity to achieve a desired placement and alignment. The control system is analyzed and proven uniformly practically asymptotically stable (UPAS). The proposed control method is validated through high-fidelity simulations that capture the intricate interaction between the USR and the surrounding fluid, and is seen to perform well in these simulations.

Sensitivity analysis of profile navigation command of underwater gliders to the initial heading error for improving trajectory accuracy

With the widespread application of underwater gliders (UGs) in marine research, there is an increasing demand for improved trajectory control accuracy. However, the initial heading error is unavoidable in internally actuated UGs, making it challenging to rectify the trajectory error generated in the heading error correction process. To effectively mitigate the impact of this error on trajectory accuracy, this study analyzes the sensitivity of profile navigation command parameters to the initial heading error through dynamic modeling of Petrel-L UG. Firstly, a simulator is developed to model the behaviors of Petrel-L, mainly involving dynamic modeling, underwater navigation, low-level control, and environment model. Subsequently, the sensitivity of profile navigation parameters in various commands to the initial heading error is analyzed. Surrogate model and Monte Carlo simulation are integrated to improve the computational speed of the analysis while preserving calculation precision. The average maximum single-profile horizontal error is 661 m, and the average maximum single-profile course error is 6.47°. Among profile navigation parameters, the target pitch during diving is proved to be the primary factor affecting the maximum trajectory error, and the increase of this parameter can enhance trajectory accuracy. This research also provides valuable reference for path following and station-keeping control of other UGs.

Iterative real-time optimization of a membrane pilot plant

In this paper, we describe the implementation and experimental validation of iterative real-time optimization (RTO) at a continuously operated membrane separation pilot plant. The objective is to achieve an economically optimal operation using the transmembrane pressure and the feed flow rate as adjustable operational parameters (set-points for the lower-level plant controllers). In spite of the popularity of RTO based on rigorous nonlinear plant models in the research literature, its practical use so far remains limited to large scale plants in the petrochemical industries. The main issue behind the limited application of RTO schemes is the necessary effort to derive and maintain sufficiently accurate models of the plants. In case of a significant plant-model mismatch, the results are not reliable, the optimal performance is not reached and even systematic violations of constraints may occur. The applicability of RTO could be extended significantly if simple process models sufficed for its successful application. Then an adaptation of the model or of the optimization problem to the real behavior of the plant is necessary to achieve an optimal operation. For this example, we tackle this issue by applying an iterative scheme called modifier-adaptation with quadratic approximation in combination with model adaptation to ensure the convergence of the plant to the true process optimum despite the fact that the model has deficiencies. The experimental results show that, in spite of a considerable plant-model mismatch, the presented RTO scheme is capable of boosting the process economics.

Adaptive patient-cooperative compliant control of lower limb rehabilitation robot

With the increase in the number of stroke patients, there is a growing demand for rehabilitation training. Robot-assisted training is expected to play a crucial role in meeting this demand. To ensure the safety and comfort of patients during rehabilitation training, it is important to have a patient-cooperative compliant control system for rehabilitation robots. In order to enhance the motion compliance of patients during rehabilitation training, a hierarchical adaptive patient-cooperative compliant control strategy that includes patient-passive exercise and patient-cooperative exercise is proposed. A low-level adaptive backstepping position controller is selected to ensure accurate tracking of the desired trajectory. At the high-level, an adaptive admittance controller is employed to plan the desired trajectory based on the interaction force between the patient and the robot. The results of the patient–robot cooperation experiment on a rehabilitation robot show a significant improvement in tracking trajectory, with a decrease of 76.45% in the dimensionless squared jerk (DSJ) and a decrease of 15.38% in the normalized root mean square deviation (NRMSD) when using the adaptive admittance controller. The proposed adaptive patient-cooperative control strategy effectively enhances the compliance of robot movements, thereby ensuring the safety and comfort of patients during rehabilitation training.

A robotic platform for integrating biomimetic sonar and AI

To gain a first-hand view of the biosonar inputs that a bat receives when operating in a complex natural environment, a bioinspired sonar head to mimic the auditory periphery of the bats' biosonar system is currently under development. The sonar head seeks to replicate the static geometric complexity of the baffles that emit the biosonar pulses and receive the returning echoes in horseshoe bats. Furthermore, some of the dynamic deformations of these structures are mimicked using tension-driven soft-robotics actuated by a set of 12 motors (two for the noseleaf and five for the each pinna). The periphery has been interfaced with acoustic emission and reception systems as well as a modular system control system for the acoustic input and output as well as the motors. Besides low-level control, the sonar head has onboard computing resources that can support deep-learning inference in the loop with acoustic data acquisition and pulse generation. All functional components have been integrated into a custom-designed shell that allows operating the system in natural outdoor environments. A wireless user interface allows remote control over the system so that the tasks of carrying and orienting the sonar head and controlling its function can be shared between two individuals.

Hierarchical Velocity Optimization for Connected Automated Vehicles With Cellular Vehicle-to-Everything Communication at Continuous Signalized Intersections

The rapid development of intelligent connected technologies and cellular vehicle-to-everything communication (C-V2X) provide new opportunities to solve the connected automated vehicle (CAV) traffic problem for eco-driving at continuous signalized intersections. With C-V2X, a hierarchical velocity optimization design based on hybrid model predictive control technique (HVO-HMPC) is presented to reduce the fuel consumption and pollution emission. First, a distance–domain velocity optimization problem, with distance as the independent variable, was constructed. Second, a hybrid MPC scheme was developed by combining the multiple shooting method and MPC technique to calculate the optimal velocity profile of a high-level controller, which acts as the reference velocity in a low-level controller. Then, a car-following model was built, the low-level controller tracked the reference velocity with the predictive control as the backbone, and the optimal velocity was calculated while ensuring that the safety velocity constraint is satisfied. Next, the proposed HVO-HMPC was tested in Prescan, and the effect comparisons with different control methods in terms of fuel consumption, pollution emission, braking time, and number of braking applications were studied under different driving scenarios. Results show that once the maximal speed is limited to 40 km/h under short-period signals and 20 km/h under long-period signals, the HVO-HMPC effectively reduces fuel consumption by 27.21%, 25.89%, and pollution emissions by 25.3%, 25.97%, respectively, while achieving best performance. Finally, an experimental prototype is built to confirm the validity of the HVO-HMPC.

Signal Compensation Control With DDPG-Based Tuning Strategy for Electronic Throttle System

The electronic throttle system (ETS) is one of the important components of an automobile engine that adjust the engine air intake to directly affect the combustion power of the engine. In practical scenarios, Proportional-Integral-Derivative (PID) is a common and effective method. However, the traditional PID is difficult to satisfy the ETS performance requirements, e.g., the system response overshoot is not allowed to exist. Moreover, the controller gains are also difficult to tune manually. To tackle the above-mentioned issues, we propose a signal compensation control framework with an adaptive gain tuning strategy for the electronic throttle system, which consists of an upper-level to tune controller gains and a lower-level signal compensation controller to eliminate the nonlinear terms. The main feature of the signal compensation controller relies on two nonlinear compensators to remove the strong nonlinearity and uncertainty in the actual throttle system. Furthermore, we develop an adaptive gain tunning strategy based on Deep Deterministic Policy Gradient (DDPG) algorithm, and we analyze the corresponding stability and convergence properties of the proposed method. Finally, we illustrate the proposed algorithm on benchmarks and the tracking control problem using the real-world experiment platform of the ETS to show the effectiveness of the proposed algorithm.

Reducing Oscillations for Obstacle Avoidance in a Dense Environment Using Deep Reinforcement Learning and Time-Derivative of an Artificial Potential Field

Obstacle avoidance plays a crucial role in ensuring the safe path planning of quadrotor unmanned aerial vehicles (QUAVs). In this study, we propose a hierarchical framework for obstacle avoidance, which combines the use of artificial potential field (APF) and deep reinforcement learning (DRL) for training low-level motion controllers. Unlike traditional potential field methods, our approach modifies the state information received by the motion controllers using the outputs of the APF path planner. Specifically, the assumed target position is pushed away from obstacles, resulting in adjustments to the perceived position errors. Additionally, we address path oscillations by incorporating the target’s velocity information, which is calculated based on the time-derivative of the repulsive force. Experimental results have validated the effectiveness of our proposed framework in avoiding collisions with obstacles and reducing oscillations.

Symmetry-Informed Reinforcement Learning and its Application to Low-Level Attitude Control of Quadrotors

Symmetry is ubiquitous in nature, physics, and mathematics. However, a classical symmetry-agnostic Reinforcement Learning (RL) approach cannot guarantee to respect symmetry. Researchers have shown that if the symmetry of a system cannot be respected, the performance of a symmetry-agnostic RL approach can be inhibited. To this end, this paper develops a generally applicable Neural Network (NN) module with symmetry that can enforce the symmetry of a system to be respected. Based on the NN module with symmetry, this paper proposes a symmetry-informed Model-Based RL (MBRL) approach that respects symmetry and improves data efficiency. The symmetry-informed MBRL approach is applied to the attitude control of a quadrotor in simulation to evaluate the effectiveness of the approach. The simulation results show that the data efficiency of the symmetry-informed MBRL approach is much superior to that of a symmetry-agnostic MBRL approach. An NN module with symmetry can respect the symmetry of a quadrotor while a naive NN cannot enforce the symmetry of a quadrotor to be respected.

A Lyapunov based posture controller for a differential drive mobile robot

Driving a vehicle to a desired position and orientation is one of the most important problems that should be solved in most navigation systems. This paper describes a new complete design and hardware implementation of a two-level controller that will enable a differential drive mobile robot to reach any desired posture starting from any initial position. The first or low-level controller consists of a set of two proportional–integral–derivative (PID) controllers, running on an embedded system on board of the robot. These controllers provide the required voltages to the motors to make the left and right wheels of the robot rotate with the angular speeds computed by the second or high-level controller, running on a stationary PC system. This second controller is based on the Lyapunov stability theorem to derive two control laws for the kinematic model, used to transform the linear and angular speeds of the unicycle model in terms of left and right rotational speeds, required by the motors. As will be shown later, this architecture provides a very flexible way not only to tune the main controller parameters but also to get access and record all the system states.

Design, Assembly and Control of a Differential/Omnidirectional Mobile Robot through Additive Manufacturing

Although additive manufacturing is a relatively new technology, it has been widely accepted by industry and academia due to the wide variety of prototypes that can be built. Furthermore, using mobile robots to carry out different tasks allows greater flexibility than using manipulator robots. In that sense, and based on those above, this article focuses on the design and assembly of a multi-configurable mobile robot that is capable of changing from a differential to an omnidirectional configuration. For this purpose, a sequential mechatronic design/control methodology was implemented to obtain an affordable platform via additive manufacturing which is easily scalable and allows the user to change from one configuration to another. As a proof of concept, this change is made manually. Fabrication, construction, and assembly processes for both structures are presented. Then, a hierarchical control law is designed. In this sense and based on Lyapunov’s method, a low-level controller is developed to control the angular speed of the wheels to a desired angular speed, and a medium-level controller controls the robot’s attitude to follow a desired Cartesian trajectory. Finally, the control strategies are implemented in both prototype configurations, and through experimental results, the theoretical analysis and the construction of the mobile robot are validated.

Recent studies on hypertension prevalence and control in India 2023.

Hypertension is the most important chronic disease risk factor in India. Recent epidemiological studies have reported that hypertension is increasing in India with a more rapid increase in rural and young populations. Fifth National Family Health Survey (NFHS-5) and Indian Council of Medical Research-INDIAB surveys have reported that there are substantial geographic variations in hypertension prevalence with greater prevalence in more developed states and districts of the country. There is a high prevalence of young-age hypertension, especially in the less developed states. The incidence of adverse events from hypertension-related cardiovascular disease is significantly greater in India than in more developed countries. A low level of hypertension awareness, treatment, and control, especially in rural and underserved urban populations is an important finding. In this narrative review, we highlight recent nationwide studies and unique features of hypertension in India and suggest strategies for better hypertension management and control.