This paper reviews recent results on the use of extended high-gain observers as disturbance estimators. It shows how the use of these observers enables recovering the performance of feedback linearization controllers. It shows also how the observer can be combined with dynamic inversion. Three experimental applications are described.
In this paper, to reduce the computation time for solving the finite-time optimal control problem (i.e., the model predictive control (MPC) problem), an approximate model of switched linear systems is proposed. A switched linear system is one of the typical subclasses of hybrid systems, and has several applications such as networked control systems. First, an approximation of the state variable in switched linear systems is derived. Next, using auxiliary binary and continuous variables, an approximate model given by a mixed logical dynamical system model is derived. Using this approximate model, the dimension of the auxiliary continuous variable can be reduced. As a result, the online computation time in MPC can decrease. Finally, the effectiveness of the proposed method is presented by a numerical example.
For robotic joints with compliant transmissions, Mahvash and Okamura have proposed a friction compensator based on an elastoplastic friction model. One drawback of the compensator is that the compensator continues producing non-zero output force in the static friction state, which results in the degraded backdrivability of joints. In order to remedy this problem, this paper proposes an elastoplastic friction compensator with improved static friction behavior, which is realized by an additional term that makes the output force exponentially decay in the static friction state. This paper also proposes an additional algorithm that adjust the decay rate of the output force in real time. The proposed methods are experimentally tested with a linear actuator with a timing belt and a six-axis industrial manipulator.
The negative-going Bereitschaftspotential (BP) is associated with the preparation and execution of dynamic movement. So far, BP for simple movements involving either the upper or the lower body has been studied. However, BP has not yet been recorded during both movements. Our study reveals that the negative-going BP was evoked around 3 to 2 seconds before the onset of the rise in response to a start cue. The BP had a negative peak before the onset of the movement of the trunk. BP for the rise to standing up was started around -3 seconds. The corresponding BP while seated could not be recorded in response to the other start cue. Then we tried to discriminate the onset for rise to stand, and for keeping seated using BP data in time domain. BP for 0.5 seconds in the time window is used for the time-shifted leave-one-out cross validation method. The method shows that using 0.5 seconds to 2 seconds before the onset of the movement could detect whether the subject would stand up or keep seated with the correction rate of about 60% to 70%. These results suggest that using the negative BP can detect the subject's will to stand-up.
In this paper, we proposed an approach that uses genetic programming (GP) to automatically develop a driving agent - as a model of a human driver - to optimally steer a realistically simulated car with an instant, non-latent steering response. Using the latter, we tested the hypothesis that introducing a delay in the steering response of an evolved model of a human driver results in well-expressed steering oscillations. We further evaluated the effects of speeding on the steering oscillation and observed that (i) the evolved model of the human driver (driving agent) could robustly control the car driven at various speeds and (ii) the increase in speed results in increase of amplitude of observed steering oscillations. The detection of these oscillations could pave the way for providing early warnings of inadequate driver cognitive load (as an underlying cause of such delays) in normal driving conditions - well before any urgent response to an eventual hazardous traffic situation may be required.
This paper introduces a new digital filter utilizing a SINC filter used in ΔΣ A/D converters and featuring 50/60Hz power-line noise rejection. Compared with a simple series connection of the SINC filter and a noise suppression filter for the power-line, this digital filter reduces the number of gates by approximately 40% and enables the attenuation of 50Hz and 60Hz elements by approximately 55dB.
In virtual reality (VR) technology, force feedback/display devices are expected to enhance fine sensation and contribute to the user's immersive experience. This is particularly true for glove-type devices. Force display gloves, however, do not have a specific numerical value to indicate display accuracy. In this paper, we report the results of sensory testing and analysis with VR cylindrical objects using a passive-type force display glove. The size discrimination threshold for grasping was determined to be smaller than the nominal average error of the device; however, the threshold was larger than either the actual accuracy of the device or the threshold from an organoleptic test of a barehanded human. We have also identified the possible superimposition of the sensory accuracy errors of the operator with those of the glove. Despite this finding, we have determined that the device shows adequate accuracy for representing cylindrical objects during specific grasping tasks.
Velocity perception is important in safety, traffic jam relaxation, and concentration and drivability enhancement while driving automobiles. This paper study the control of velocity perception by using vibrotactile stimulation. We have focused on vibrotactile stimulation since tactile stimulation can be reasonable and be privately given to a driver. First, the seat vibration is measured using an actual vehicle to discover available stimulation of changing velocity perception. The result shows the peak frequency clearly increases with the rise of velocity while the intensity roughly increases. Then we investigate the influence of intensity or frequency-shifted vibrotactile stimulation on velocity perception by using a driving simulator reflecting actual vehicle characteristics. An experimental result shows a possibility of changing velocity perception by using frequency-shifted vibrotactile stimulation.
This paper proposes a method that aggregates similar portions (code clones) in the developed program and redesigns the current program structure to an adequate program structure. The procedure of the proposed method is the following: Code clones are detected, and their characteristics are extracted, the design methods that code clones are aggregated are proposed, the program is modified, and the affected portions in the program are clarified and retested. Furthermore, a tool that supports the proposed method is created. As a result of applying the proposed method and the tool, it is found that 45% CCs are aggregated and 13% instructions of the program are deleted. Additionally, it is found that an adequate program structure is realized by using the proposed method and the tool.
Interplanetary micro-spacecraft have been recently demonstrated for various missions. Orbital control of micro-spacecraft is challenging because the spacecraft systems have severe constraints. This paper presents a constrained optimal flyby guidance algorithm. We iteratively introduce two-stage stochastic programming and achieve rapid and global optimization by using delta-v mappings and cross-correlation technique, which is well-known in the field of image and signal processing. Finally, numerical examples and Monte-Carlo simulation show that the proposed algorithm is efficient in terms of the expected total delta-v.
We investigate the relationship between graph topology and delay-independent synchronization occuring in networks of the identical nonlinear systems with transmission delays. In this paper, we show that if networks contain a cycle subgraph of an odd number of nodes, partial synchronization corresponding to the equitable graph partition with the fewest cells occurs for a sufficiently large coupling strength regardless of the length of time-delay. The validity of the obtained results are examined through numerical simulations of Hindmarsh-Rose neuron systems networks.
This paper studies visual feedback position tracking control of two-wheeled vehicles in the situation that a camera and a target object are attached to the vehicles, respectively. Here, the body velocity of the target object vehicle is modeled as a Fourier series expansion. The relative position between the camera and the target object is controlled to the desired relative position. The present control law is based only on visual measurements, and the necessary information to implement the law is estimated from them. The asymptotic stability of the equilibrium of the total system including the internal attitude behavior is analyzed under some conditions of the target object vehicle velocity via stability theory of perturbed systems. Finally, a numerical simulation is conducted to show the effectiveness of the proposed method.
Recently, increasing attention has been paid to the health impacts and lifestyle diseases due to the excessive intake of salt and sugar. To address this problem, there have been studies to control taste by the electrical stimulation of food and drink through tableware. However, it is not possible to realize a change of taste during the chewing food and drinking with this approach. Therefore, we consider that a change of taste is also required during chewing in order to achieve satisfied consumption. In this study, we develop an intra-oral interface to effect a taste change continuously by using electrical stimulation. We performed preliminary experiments to measure the sensitivity of the tongue by the position and area of an electrode. Then, we developed an interface to take into account the results of the experiments. We confirmed the effectiveness of the proposed interface through experiments examining the intensity of taste when using the developed interface. Moreover, by having periodic intervals to electrical stimulation by the PWM control, the proposed interface was confirmed to improve the intensity of taste continuously.
This paper examines the effect of instructions on parts' positions during an assembly task having numerous parts for someone to remember where they are located. Hypothesizing that instructions relating to the positions of the parts would enhance work efficiency and reduce workload, we designed an experimental method to test this theory. In this experiment using educational blocks, visual instructions were given by illuminating the space where the parts are kept, while auditory instructions were provided to aid locating the parts to compare the efficiency of visual and auditory instructions. Experiments with six participants showed that the visual instruction significantly shortened not only the search time but also the assembly time (which was one way in which efficiency was assessed in this study). The use of parts' positions instructions tended to reduce the workload as evaluated with NASA-TLX compared with the cases without instructions.
Space debris became one of the difficulties for mankind to utilize the space environment. Among the variable efforts to the debris problem, the development of technology for the active debris removal (ADR) is especially imperative. This paper is on the motion simulation of debris capture manipulator for ADR with large number of links. A lightweight actuator using hydrogen storage alloy is developed for use in outer space as an actuator for the manipulator. The controllable time with the extended resolved motion rate control was examined in the past study. In this paper, to research the motion of the debris in the interaction with the end-effector, a simple collision model for the special end-effector was established. The control law to lower the relative velocity after the impact is brought in, and the results have shown almost 90% of reduction regardless of the initial velocity of moving target debris. The proposed system will enable the system to capture the target debris more safely.
In this paper, we establish the dynamics of a spacecraft with Variable-Speed Control Moment Gyros (VSCMGs). Based on this dynamics, we develop an easy-to-use Linear Parameter-Varying (LPV) model, in which we deal with 3-axis attitude control of a spacecraft with pyramid-array VSCMGs. Although it is hard to consider overall stability and control performance of attitude control at the same time with conventional Lyapunov function-based controllers, this paper attains both of them via LPV control theory. And also we design two types of Gain-Scheduled (GS) steering laws for singularity avoidance and escape. Through numerical examples, the efficiency of the proposed GS controller and the GS steering laws is demonstrated.
This paper proposes a control method of the magnetorheological (MR) damper semi-active suspension controlling system with an L1 adaptive controller. The suspension system can be represented as a non-homogeneous system. Therefore, the proposed controller design method is introduced for avoiding the solving process of the convolution of a transfer matrix in the non-homogeneous system and guaranteeing the performance of the system within an acceptable level. The result, which is simulated by MATLAB/SIMULINK, proves that the proposed control architecture can work well for reducing calculation process. Moreover, this method is compared with the auto-tuned PID function in SIMULINK; as a result, the method shows a better overall performance by adjusting a few parameters.
The multi-agent surveillance problem over graphs is to find trajectories of multiple agents that travel each node as evenly as possible. This problem has several applications such as city safety management and disaster rescue. In this paper, the finite-time optimal surveillance problem is formulated, and is reduced to a mixed integer linear programming (MILP) problem. Based on the policy of model predictive control, an optimal trajectory is generated by solving the MILP problem at each discrete time, and persistent surveillance can be realized. Finally, a numerical example is presented, and further discussion is also provided.
In a standard setup of conventional state estimation problems, the output signal of a dynamical system is sampled at every regular time interval. This paper addresses an estimation problem under event-triggered sampling: an irregular sampling rule in which the output signal is sampled only when a specified event occurs. In particular, this paper focuses on the Lebesgue sampling, which is a type of event-triggered sampling such that the output signal is sampled only when it crosses specific thresholds. In the proposed estimation method, not only information in the sampled points but also that in the inter-sample intervals are utilized to improve the estimation accuracy efficiently. The particle filter is applied to the estimation since the inter-sample information makes the distribution of the estimates non-Gaussian. Furthermore, the proposed estimator is combined with a linear-quadratic-regulator (LQR) type state-feedback controller. The derived control law can be an effective approach to an event-triggered control problem in which the Lebesgue-sampled output is utilized. The effectiveness of both of estimation and control methods is examined through numerical examples.
Opinion formation is investigated over signed gossip networks, where asymptotic and transient opinion behaviors are analyzed in detail. The network is modeled by a connected signed graph, where each edge of the graph has positive or negative weight, i.e., these edges represent friendly or hostile relationship. Interactions each of which occurs between two persons chosen randomly follow an asynchronous procedure. Then it is shown that their opinions always converge in mean square and converge almost surely. A convergence time is defined as the smallest time such that it takes for the values of nodes to get within a given neighborhood of the consensus value with high probability, regardless of initial state. An upper bound and a lower bound of the convergence time of the gossip algorithm are given in terms of a structure of the given network.