In this paper, we propose a workspace force feedback controller for parallel-link robots. In such a controller, the thrust reference is generated using workspace coordinate information. Therefore, forward kinematics, which calculates end-effector motion information from actuator motion, is essential. For parallel-link robots, however, forward kinematics cannot be solved algebraically and is typically computed using numerical methods. However, these methods tend to be computationally demanding. To address this issue, we propose a forward kinematics method based on an extended Kalman filter (FKbEKF). The workspace force feedback controller is constructed using this forward kinematics approach. The effectiveness of the proposed method is validated through simulation.

In this paper, we propose a motion trajectory correction method for a robot arm based on force information and a deep learning model. In recent years, the automation of manufacturing processes using robots has been progressing. However, automating manual tasks that require human technical skills, especially "craftsmanship" tasks such as application tasks, remains difficult. In contact-type application tasks using tools such as sponges, the precision of contact pressure control affects product quality. Therefore, high-precision force control is required for such tasks. In this study, we develop an automatic application system that imitates human application motions by equipping a robot arm and teaching device with force sensors. In our previous studies, the motion trajectory of the robot arm was corrected with iterative learning. However, this approach requires considerable time for correction whenever a new motion trajectory is implemented on the robot arm. Therefore, we propose a method to estimate the correction amount for the motion trajectory using a deep learning model. The proposed method is expected to significantly reduce correction time compared to our previous method.

This study addresses the challenge of hand rehabilitation, especially for those with neuromuscular impairments, by proposing a tendon-driven soft hand exoskeleton with a robust control scheme. We introduce a novel non-singular terminal sliding mode (NSTSM) control approach tailored for tendon-driven hand exoskeletons. This control strategy enhances trajectory tracking during rehabilitation exercises by addressing uncertainties and external disturbances, ensuring robust performance while avoiding singularities. Through numerical simulations, we evaluate the efficacy of the NSTSM control in accurately following desired trajectories across multiple finger joints, emphasizing singularity-free control and stability analysis based on Lyapunov theory. Our findings reveal that the NSTSM control achieves superior trajectory adherence compared to conventional PID controls, with reduced RMS error values averaging 0.005 [rad] for MCP and PIP joints, and 0.02 [rad] for DIP joint. The sliding surface analysis confirms the NSTSM control’s robustness, highlighting the NSTSM control’s consistent performance, and offering a promising approach to enhance rehabilitation outcomes in stroke survivors with hand impairments.

Carbon fiber reinforced polymer (CFRP) materials are widely used in the aerospace industry. Repair of CFRP materials entails a process referred to as scarf sanding wherein damaged areas are milled and removed. However, this process generates substantial amounts of carbon dust. Therefore, to ensure explosion-proof operation, motors operating without electrical power, such as pneumatic motors, must be utilized for the automation of scarf sanding. Additionally, angular velocitysensorless control of the pneumatic motor is essential. In this paper, we propose an angular velocity estimation method based on the mathematical model of a pneumatic motor. Further, we propose a new method for acquiring the angular velocity of a pneumatic motor based on the fast Fourier transform (FFT) of the measurements from a volume flow rate sensor installed in the motor air supply piping. Finally, we integrate both methods to devise a hybrid angular velocity estimation method. The verification results demonstrate that the hybrid method can accurately estimate the angular velocity in the range of 0–1600 RPM.

In this paper, we propose a trajectory correction method that utilizes force and bidirectional long short-term memory (BiLSTM) to reproduce the delicate glue-application task performed by a human on a robot. In the system we have developed, the operator initially performs the glue-application task using a teaching device. The trajectory and force are recorded during the task. If the robot follows the recorded trajectory, it can be expected to perform the same glue-application task as the human. However, large errors in the force occur owing to the deflection of the robot and individual differences in the application tool. To minimize this problem, our previous study used force and iterative learning to correct the trajectory. Although the error was reduced, running the robot multiple times for iterative learning took a long time. The more types of teaching are learned, the longer time is required. To solve this problem, BiLSTM was used in this study. To make BiLSTM learn correction equivalent to iterative learning, the trajectory and force recorded by the teaching device and robot during the glue-application task were used. Achieving this should allow deep learning to correct the robot’s trajectory, even for an unknown teaching move, because of its better generalization performance compared with other machine learning algorithms. In this study, we investigated whether this is possible.

This study examined the method to simultaneously conduct identical experimental classes across different locations, particularly considering the potential for remote classes during pandemics such as COVID-19. The goal was to achieve a system configuration that allows for distributed classes while maintaining a sense of unity, similar to in-person sessions. Hierarchical optimal control methods were applied for this purpose. Further-more, this study addresses the communication delays that occur during remote control between distant locations by applying the concept of state-prediction control in time-delay systems. This approach demonstrated the ability to calculate the current state of a counterpart from the past states, thereby mitigating the impact of delays. The effectiveness of this approach was demonstrated experimentally using an inverted penduluman iconic system used in control engineering education.

This paper proposes a force-controlled precision drilling process based on a reaction force estimation observer to realize precision drilling, as typified by rivet holes in aircraft, using an industrial robot arm. So far, precision drilling of aircraft panels has been performed by skilled technicians using their hands. Moreover, some precision drilling processes are performed using large and expensive special-purpose drilling machines. However, owing to their high prices and poor versatility, a small industrial robot arm is often required to realize this process. Here, we propose a high-precision drilling process that uses force control to control the thrust force of the drill on an end effector attached to a small robot arm. The effectiveness of the proposed method is verified based on actual machine verification.

Recently, geared motors, motors with which reduction gears and encoders are integrated, are developed and are becoming common. If a reduction gear and an encoder are integrated with a motor, it is possible to arrange the encoder on the load side instead of the motor side and to control the load-side angle with the full-closed loop control. The paper proposes the motor-side angle estimation for the two-mass system with a load-side encoder to realize the full-closed loop control without a motor-side encoder. The paper defines the non-linear state-space equations by considering not only the dynamics model of the two-mass system but also the voltage equation based on the power electronics region. The proposed method is designed based on the non-linear state-space equation and expanded Kalman filter, one of the non-linear Kalman filters. The paper shows the effectiveness of the proposed method with the simulation results.

The paper proposes the parameter optimization of image processing for contamination inspection of nonwoven fabrics. Currently, the automation of contamination inspection using image processing systems is being considered. In image processing, it is important to set the optimal parameters for the processing. However, it is necessary to search it from many combinations because there are some parameters. The proposed method searches for the optimal parameters based on a genetic algorithm. It reduces the search time in comparison with the conventional method. The paper indicates the effectiveness of the proposed method with the experimental results.

This study proposes a method for correcting trajectory data of a robot using iterative learning to mimic the motion of a human with a comparable force level. Currently, trajectory generation using direct teaching is often used to teach robots more flexible movements. Although such a teaching method can produce ideal trajectories, it may neglect the reproducibility of the force. Therefore, we propose a method that can reproduce the originally required sense of force by correcting the original trajectory data using iterative learning. Moreover, the reduction in the force error using our method is verified.

In order to pick bolts stacked in bulk using a robot, it is necessary to obtain the most suitable grasping position. Recently, deep learning-based methods have been proposed. However, learning takes a long time and much effort to prepare the training. In order to avoid this, this research proposes a template matching-based method. First of all, we defined some states where the bolt can be grasped stably. By efficiently searching the grasping position where the objective function is maximized using differential evolution (DE), the proposed method can acquire the grasping position in two seconds. After an original gripper was made using a 3D printer and attached to the robot, we experimented using 60 bolts (M6-20). The results show 86% of the success rate.

Snow and Ice is considered one of the disasters. As one example of efforts for disaster prevention and mitigation for expressways, we report on the safety measures against Snow and Ice countermeasures for safe and secure Tohoku-Expressway. ICT is a necessary technique for a safe and secure expressways.

This paper presents a high precision modeling approach for a multi-axis robot considering interference force between axes based on a dynamic model of a multi-axis robot. It is well-known that interference force affects positioning performance especially in high speed positioning motion. In order to overcome the problem, it needs a high precision model which can reproduce actual motion of a multi-axis robot. In this paper, a combined modeling approach which consists of a two-mass model and a dynamic model is proposed, where each axis is basically modeled as a two-mass model to reproduce mechanical vibrations due to elastic mechanism of the reduction gear. Interference force between axes is theoretically modeled based on a dynamic model of a multi-axis robot. The inertial term of interference force is added as disturbance at motor-side, while the velocity term of interference force and gravity force are added as disturbance at load-side. The effectiveness of the proposed modeling approach has been verified based on a comparative study with a conventional model approach based on a two-mass model only.

Fast and precise positioning of a multi-axis industrial robot is important solved problems in order to achieve higher productivity and cost saving. In this paper, the state feedback control approach is proposed in order to reduce mechanical vibration of the robot arm with posture changes of the arm. The control system consists of a variable 2DOF controller and a state feedback of the estimated load acceleration. In the variable 2DOF controller, the feedforward compensators are designed based on a coprime factorization description for the plant system, where the model parameters in the compensators are varied corresponding with posture change of the arm. In the state feedback, in order to avoid additional cost introducing the accelerometer to the load, the load acceleration is estimated by the variable state observer, where the plant parameters in the state observer are varied corresponding with the posture change of the arm to fix the estimation performance. The effectiveness of the proposed controller has been verified by the experiments using a prototype.

This paper presents a use of MEMS accelerometers for performance improvement of motion control systems equipped with low resolution position sensors. It is well-known that control performance depends on hardware performance such as a sensor resolution and a computational power. Above all, the use of low resolution sensors restricts the achievable enhancements on the control bandwidth, due to quantization noise. In this paper it is proposed to overcome such limitation by using a Kalman estimator, utilizing the measurements provided by a low cost MEMS accelerometer in the estimation of position information. Moreover, the same measurement is used for the acceleration based disturbance observer (a-DOB), where no differentiator is needed to estimate the disturbance, so that a better estimation capability and positioning performance can be achieved. The effectiveness of the proposed control system has been verified by several experiments using a prototype of a table system.

Recent years, real world haptic techniques have been developed. A bilateral control scheme is utilized for haptic devices, where force and/or position information of master slave system is transported each other. It is well-known that control performance of bilateral systems strongly depends on hardware restriction and computational power, however providing high performance hardware needs additional cost. In this research, a steer-by-wire(SBW) system is utilizing as one example of the bilateral control system with low resolution position sensors. For this system, in order to improve estimation capability, a state observer utilized for interpolation of position information is applied to the reaction force observer. Moreover in order to solve the problem that non-linear friction in ball guide of the slave system afl'ects the positioning performance of the system, model-based feed forward disturbance compensation is applied to compensate for non-linear friction. The effectiveness of the proposed method has been verified by experiments using a prototype of the SBW system.