Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (scientific journal)   Publisher：Institute of Electrical and Electronics Engineers (IEEE)  

DOI： 10.1109/tmech.2016.2614961

Scopus

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (scientific journal)   Publisher：SAGE Publications  

Our paper proposes a tendon-driven robotic finger based on an anatomical model of a human finger and a suitable method for its analysis. Our study aims to realize an anthropomorphic robotic hand that has the same characteristics and dexterity as that of a human hand, and it also aims to identify the advantages of the human musculoskeletal structure for application to the design and control of robot manipulators. When designing an anthropomorphic robotic hand, several devices are required to apply the human finger structure to a tendon-driven robotic finger. Reasons for this include that one of the human finger muscles, namely, the lumbrical muscle, is situated between tendons, which is an unfavorable configuration for the tendon-driven mechanism. Second, unlike a standard pulley used in a tendon-driven mechanism, some moment arms of the human finger change nonlinearly according to the joint angle. In our robotic finger design, we address these difficulties by rearranging its tendons and develop a mechanism to change the moment arm. We also propose a method to analyze and control this robotic fingers coordinating joints using non-stretch branching tendons based on the human extensor mechanism with a virtual tendon Jacobian matrix and the advantage is that this constraint virtually reduces the degrees-of-freedom (DOF) of the mechanism. Further, we build a prototype to confirm its motion using this method. In addition, we show that the state with the reduced DOF can be lost by external forces acting on the mechanism, and this condition can be changed manually by adjusting the tendon forces. This makes it possible to control the virtual DOFs to satisfy the requirements of the task. Finally, we discuss the benefits from anthropomorphic structures including the tendon arrangement, which mimic the human lumbrical muscle, and the above mentioned mechanism with non-linear moment arms from the perspective that there are two states of DOFs. These insights may provide new perspectives in the design of robotic hands.

DOI： 10.1177/0278364913518357

Scopus

researchmap

Other Link： http://journals.sagepub.com/doi/full-xml/10.1177/0278364913518357

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (scientific journal)   Publisher：Elsevier BV  

DOI： 10.1016/j.robot.2012.07.016

Scopus

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/mhs59931.2023.10510021

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：ASME International  

Abstract

The transportation of large-scale objects in a narrow space is a challenging, but useful application for mobile robots. We have developed a three-mobile-robot system to cooperatively lift, support, and transport a large object on a mobile robot system. To facilitate the manipulation involved with loading an object onto the robots, where the robots must maintain firm contact with the object and anchor at its location, we designed an adaptable mechanism for the object-loading platform and a liftable brake to ensure that the robot remains stationary when necessary. Furthermore, each robot is designed to act as an omnidirectional wheel; therefore, all three robots can work as an omnidirectional block when collectively loading an object. The kinematic constraints of the object–robot system and forward kinematics of the robots’ cooperative motion are proposed, and experiments are conducted to confirm the designed mechanisms and confirm that the robots can load an object and cooperatively transport it along the expected trajectory.

DOI： 10.1115/1.4056771

researchmap

Authorship：Lead author   Language：English   Publishing type：Research paper (scientific journal)   Publisher：Institute of Electrical and Electronics Engineers (IEEE)  

DOI： 10.1109/jsen.2022.3200017

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Oxford University Press (OUP)  

Abstract

The Fugl-Meyer Assessment is widely used to test motor function in stroke survivors. In the Fugl-Meyer Assessment, stroke survivors perform several movement tasks and clinicians subjectively rate the performance of each task item. The individual task items in the Fugl-Meyer Assessment are selected on the basis of clinical experience, and their physiological relevance has not yet been evaluated. In the present study, we aimed to objectively rate the performance of task items by measuring the muscle activity of 41 muscles from the upper body while stroke survivors and healthy participants performed 37 Fugl-Meyer Assessment upper extremity task items. We used muscle synergy analysis to compare muscle activity between subjects and found that 13 muscle synergies in the healthy participants (which we defined as standard synergies) were able to reconstruct all of the muscle activity in the Fugl-Meyer Assessment. Among the standard synergies, synergies involving the upper arms, forearms and fingers were activated to varying degrees during different task items. In contrast, synergies involving posterior trunk muscles were activated during all tasks, which suggests the importance of posterior trunk muscle synergies throughout all sequences. Furthermore, we noted the inactivation of posterior trunk muscle synergies in stroke survivors with severe but not mild impairments, suggesting that lower trunk stability and the underlying activity of posterior trunk muscle synergies may have a strong influence on stroke severity and recovery. By comparing the synergies of stroke survivors with standard synergies, we also revealed that some synergies in stroke survivors corresponded to merged standard synergies; the merging rate increased with the impairment of stroke survivors. Moreover, the degrees of severity-dependent changes in the merging rate (the merging rate–severity relationship) were different among different task items. This relationship was significant for 26 task items only and not for the other 11 task items. Because muscle synergy analysis evaluates coordinated muscle activities, this different dependency suggests that these 26 task items are appropriate for evaluating muscle coordination and the extent of its impairment in stroke survivors. Overall, we conclude that the Fugl-Meyer Assessment reflects physiological function and muscle coordination impairment and suggest that it could be performed using a subset of the 37 task items.

DOI： 10.1093/braincomms/fcac200

researchmap

Other Link： https://academic.oup.com/braincomms/article-pdf/4/4/fcac200/45407952/fcac200.pdf

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Editorial Committee of Synthesiology  

DOI： 10.5571/synth.2021.2_1

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Institute of Electrical and Electronics Engineers (IEEE)  

DOI： 10.1109/tpami.2020.2963980

Scopus

PubMed

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Elsevier BV  

DOI： 10.2139/ssrn.3859019

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/biorob49111.2020.9224321

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/iros45743.2020.9341694

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Springer Science and Business Media LLC  

This paper proposes a system for tracking human position and head direction using fisheye camera mounted to the ceiling. This is believed to be the first system to estimate head direction from ceiling-mounted fisheye camera. Fisheye histograms of oriented gradients descriptor is developed as a substitute to the histograms of oriented gradients descriptor which has been widely used for human detection in perspective camera. Human body and head are detected by the proposed descriptor and tracked to extract head area for direction estimation. Direction estimation using randomized ferns is adapted to work with fisheye images by using the proposed descriptor, guided by the direction of movement. With experiments on available dataset and new dataset with ground truth, the direction can be estimated with average error below 40 circle, with head position error half of the head size.

DOI： 10.1007/s00371-019-01749-9

Web of Science

Scopus

researchmap

Other Link： http://link.springer.com/article/10.1007/s00371-019-01749-9/fulltext.html

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Informa UK Limited  

A combination of fixed fisheye camera system and a blimp performed the task of human following in an indoor environment in this paper, focusing on tracking the subject's face. Hierarchical control algorithm is used to combine information from fisheye cameras and blimp's camera to track the face to handle common human's translation and rotation movements. In this way, even when the face is not visible in the blimp's camera, the relative position between the blimp and the person is still available from fisheye cameras. The blimp can be controlled to move closer to the face, where it should be detectable. If the face can be detected and matched with the position of the person, the blimp can use it to adjust its position for a better track of the face. An experiment was conducted inside an office environment to demonstrate human following behavior of the blimp. With the face invisible to the blimp's camera 34.6 % of the experiment time, it could still move to recover the track of the face, using information from the fisheye cameras.

DOI： 10.1080/01691864.2020.1747537

Web of Science

Scopus

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：MDPI AG  

Muscle functional MRI (mfMRI) is an imaging technique that assess muscles’ activity, exploiting a shift in the T2-relaxation time between resting and active state on muscles. It is accompanied by the use of electromyography (EMG) to have a better understanding of the muscle electrophysiology; however, a technique merging MRI and EMG information has not been defined yet. In this paper, we present an anatomical and quantitative evaluation of a method our group recently introduced to quantify its validity in terms of muscle pattern estimation for four subjects during four isometric tasks. Muscle activation pattern are estimated using a resistive network to model the morphology in the MRI. An inverse problem is solved from sEMG data to assess muscle activation. The results have been validated with a comparison with physiological information and with the fitting on the electrodes space. On average, over 90% of the input sEMG information was able to be explained with the estimated muscle patterns. There is a match with anatomical information, even if a strong subjectivity is observed among subjects. With this paper we want to proof the method’s validity showing its potential in diagnostic and rehabilitation fields.

DOI： 10.3390/s20030724

Scopus

PubMed

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Elsevier BV  

DOI： 10.1016/j.promfg.2020.04.025

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/embc.2019.8856616

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/icorr.2019.8779462

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：MDPI AG  

If we define a mode as a set of specific configurations that hold the same constraint, and if we investigate their transitions beforehand, we can efficiently probe the configuration space by using a manipulation planner. However, when multiple mobile robots together manipulate an object by using the non-prehensile method, the candidates for the modes and their transitions become enormous because of the numerous contacts among the object, the environment, and the robots. In some cases, the constraints on the object, which include a combination of robot contacts and environmental contacts, are incapable of guaranteeing the object’s stability. Furthermore, some transitions cannot appear because of geometrical and functional restrictions of the robots. Therefore, in this paper, we propose a method to narrow down the possible modes and transitions between modes by excluding the impossible modes and transitions from the viewpoint of statics, kinematics, and geometry. We first generated modes that described an object’s contact set from the robots and the environment while ignoring their exact configurations. Each multi-contact set exerted by the robots and the environment satisfied the condition necessary for the force closure on the object along with gravity. Second, we listed every possible transition between the modes by determining whether or not the given robot could actively change the contacts with geometrical feasibility. Finally, we performed two simulations to validate our method on specific manipulation tasks. Our method can be used in various cases of non-prehensile manipulations by using mobile robots. The mode transition graph generated by our method was used to efficiently sequence the manipulation actions before deciding the detailed configuration planning.

DOI： 10.3390/app9030462

Scopus

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：MDPI AG  

Local image features are invariant to in-plane rotations and robust to minor viewpoint changes. However, the current detectors and descriptors for local image features fail to accommodate out-of-plane rotations larger than 25°–30°. Invariance to such viewpoint changes is essential for numerous applications, including wide baseline matching, 6D pose estimation, and object reconstruction. In this study, we present a general embedding that wraps a detector/descriptor pair in order to increase viewpoint invariance by exploiting input depth maps. The proposed embedding locates smooth surfaces within the input RGB-D images and projects them into a viewpoint invariant representation, enabling the detection and description of more viewpoint invariant features. Our embedding can be utilized with different combinations of descriptor/detector pairs, according to the desired application. Using synthetic and real-world objects, we evaluated the viewpoint invariance of various detectors and descriptors, for both standalone and embedded approaches. While standalone local image features fail to accommodate average viewpoint changes beyond 33.3°, our proposed embedding boosted the viewpoint invariance to different levels, depending on the scene geometry. Objects with distinct surface discontinuities were on average invariant up to 52.8°, and the overall average for all evaluated datasets was 45.4°. Similarly, out of a total of 140 combinations involving 20 local image features and various objects with distinct surface discontinuities, only a single standalone local image feature exceeded the goal of 60° viewpoint difference in just two combinations, as compared with 19 different local image features succeeding in 73 combinations when wrapped in the proposed embedding. Furthermore, the proposed approach operates robustly in the presence of input depth noise, even that of low-cost commodity depth sensors, and well beyond.

DOI： 10.3390/s19020291

Scopus

PubMed

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Elsevier BV  

DOI： 10.1016/j.promfg.2019.03.057

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-030-01370-7_31

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-030-01370-7_14

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/mhs.2018.8887058

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/robio.2018.8665241

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Fuji Technology Press Ltd.  

A method for modifying robot behaviors is introduced to improve robot performance during the execution of object manipulation tasks. The purpose of this method is to minimize the execution time of tasks and prevent collision with obstacles, including objects to be manipulated and the robot itself, by considering two approaches. The first is to use the potential that robots can provide, considering that the programs are based on events that are subject to the response of sensors. The second is to determine the maximum rate at which commands can be sent, without affecting the responses from the sensors, and, based on that, to accelerate or decelerate the execution of the task. The proposed method focuses on the refinement of two approaches: (a) modifying the trajectory of some behaviors, so that they are not executed step by step, but are executed in parallel, and (b) increasing the rate of sending robotic commands. To validate the proposed method, four real-world tasks are presented, including the flipping of a briefcase, the flipping of a weighing scale, the lifting of a weighing scale, and the opening of a folding chair, performed by a set of small robots. The reduction in execution time of the tasks varied between 54.2% and 73.6%; the implications of the improvement are discussed based on experimental results.

DOI： 10.20965/jrm.2018.p0613

Web of Science

Scopus

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/mhs.2017.8305265

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/robio.2017.8324413

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Fuji Technology Press Ltd.  

[abstFig src='/00290002/14.jpg' width='300' text='Two robots performing the task of opening the folding chair lying on the floor' ] In this study, we present a novel framework to address the problem of teaching manipulation tasks performed by a single human to a set of multiple small robots in a short period. First, we focused on classifying the manipulation style used during a human-performed task. An allocator process is proposed to determine the type and number of robots to be taught based on the capabilities of available robots. Then, according to the detected task requirements, robot behaviors are generated to create robot programs by splitting human demonstration data. Small robots were used to evaluate our approach in four defined tasks that were taught by a single human. Experiments demonstrated the efficiency of the method to classify and judge whether the division of a task is necessary or not. Moreover, robot programs were generated for manipulating selected objects either individually or in a cooperative manner.

DOI： 10.20965/jrm.2017.p0419

Scopus

CiNii Books

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-319-48036-7_4

researchmap

Authorship：Lead author,　Corresponding author   Publishing type：Research paper (international conference proceedings)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-319-48036-7_58

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/robio.2016.7866365

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/mhs.2016.7824149

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-319-08338-4_84

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/robio.2015.7418962

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/iros.2014.6943108

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Part of collection (book)   Publisher：Springer International Publishing  

DOI： 10.1007/978-3-319-09435-9_25

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/robio.2012.6490995

researchmap

Authorship：Lead author,　Corresponding author   Language：English  

researchmap

Authorship：Lead author,　Corresponding author   Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

In dexterous robotic manipulation it is essential to control the force exerted by the robot hands while grasping. This paper describes a method by which robot hands can be controlled on the basis of previous experience of slippage of objects held by the hand. We developed an anthropomorphic human scale robot hand equipped with an elastic skin in which two types of sensors are randomly embedded. One of these sensors is a piezoelectric polyvinylidenefluoride (PVDF) film which can be used for the detection of pressure changes. The other is a strain gauge which can measure static pressure. In our system, PDVF films are used to detect slipping, and strain gauges to measure stresses which are caused by normal and shear forces. The stress measured by the strain gauges are used as input data to the neural network which controls the actuators of the robot. Once a slip is detected, the neural network is updated. We show that this system can control the grasp force of the robot hand and adapt it to the weight of the held object. By using this method, it wa shown that robots can hold grasped objects safely. © 2011 IEEE.

DOI： 10.1109/icar.2011.6088598

Scopus

researchmap

Event date： 2018

Language：Japanese  

<p>Techniques called "Fitting", which considers the individual swing characteristics and suggests a golf club optimal for the player, have attracted attention. In the system developed by the authors so far, we can find optimally designed the golf clubs for the individual swing by swinging nine clubs of different characteristics. In this study, for researching relationship with golf swings and club characteristic, we measured four golfer's swings by motion capture system. Subjects swung two clubs that differ the shaft kick point, and we composed observance matrix from positional data. We conducted SVD to this observance matrix, and conducted Three-way ANOVA to each club motion. As a result, we can decompose golf swing into simple motion combination, and clarify the timing to be affected by the difference in shaft kick point.</p>

researchmap

Event date： 2018

Language：Japanese  

<p>In this paper, we dealt with the problem of designing of a serial manipulator which realize the given trajectory of its end-effector approximately with a small number of joints. Most of the cases, there is no analytical solution for this kind of problems. We proposed the methodology to solve this problem using the optimization with generalized inverse kinematics.</p>

researchmap

Event date： 2017

Language：Japanese  

<p>Non-circular pulleys and a wire can be used to constrain the motion between two rotational joints in order to generate a given coordinated motion of the joints. In this paper, the methodology to derive the shapes of non-circular pulleys for a given coordinated motion of the joints is introduced. Furthermore, the shapes of non-circular pulleys to move the robotic leg with four joints using the wire constraints are derived by this methodology for both of the stance and swing phase. The designed robotic leg can be used to develop a new passive dynamic walker.</p>

researchmap

Event date： 2017

Language：Japanese  

<p>Manipulation of heavy objects using traditional mobile robots is very difficult because they often fall by reactive forces when they apply a large force to an object. In this study, small mobile robots with a linear actuator and a passive joint at the foot of it are developed not to fall by these reactive forces. The robots can variously manipulate heavy objects by cooperating with one another. As an example, inclination of a heavy object, which is important to realize transportation, is demonstrated by the cooperation of the two developed robots. They succeeded to incline an object more than 40kg.</p>

researchmap

Event date： 2017

Language：Japanese  

<p>A flat belt is widely used in various mechanical devices to transmit a force. In this paper, a new locking mechanism to lock a flat belt using frictional forces between the flat belt and the curved surface is proposed based on the novel way to analyze the statics of an overlapping flat belt. The unlocking mechanism which changes the contact region of the flat belt with a small force is also proposed. Using this mechanism, the flat belt can be switched between being locked and unlocked.</p>

researchmap

Event date： 2016

Language：Japanese  

小型移動ロボットにより大型物体のマニピュレーションをするには，作業反力によるロボットの転倒，すべりという問題を解決する必要がある．本研究では，傾け作業におけるロボットの転倒の条件の整理，ロボットと床面，作業対象物間での摩擦の解析により，ロボットが転倒，すべりを起こさずに可能な限り力を発揮できる機構を提案した．また，実際に機構を制作し，ロボットが自律的にマニピュレーションを行うシステムを構築した．

researchmap

Event date： 2016

Language：Japanese  

<p>Human can pinch and grasp an object using his finger or fingertip with multiple contact points. The multiple-point contact increases stability while the object is grasped and controlled. We developed the anthropomorphic robotic finger with non-elastic branching tendon mechanism in order to realise the reaching motion with less actuators in our previous work. However, it was difficult to accomplish the grasping with multipoint contact. Thus, we develop a new robot hand system equipped branching tendons which realizes multiple-point contact while it grasps the object using with the whole fingers or finger-tips. In this paper, we introduce the developed robotic hand and show the reason why the robotic hand can realize the multiple-point contact.</p>

researchmap

Event date： 2014

Language：Japanese  

In tendon-driven mechanism, the relation between the motion of each actuator and each joint is complex because of its dependency on its tendon routing and moment arms. We propose the mechanism to solve this complexity composed of step pulleys and moving pulleys, and we call it as tendon inverse mechanism. Furthermore, we can control the joint stiffness independently by using non-lenear elastic components because of this mechanism. In this paper, we introduce the methodology to construct this mechanism and validate it through some experiments.

researchmap

Event date： 2014

Language：Japanese  

The most important goal for developing robotic hands is realizing stability of the grasping and dexterous manipulation with a simple control law. To achieve them, we focus on a branching tendon mechanism in a human finger. Links connected with a nonstretch branching tendon are coupled by a geometric restriction. The coupling disappears when one of branching tendon slacks. By utilizing these features, the hand may be able to grasp with a simple control and manipulate the grasped object. We developed the robotic finger with two degrees of freedom and two tendons. We proposed a joint angular position control for this finger. We derived the condition for a branching tendon to slack, and validated it by an experiment. Experimental results demonstrated that the hand can be position-controlled when it moved freely, and that the coordination between joints changed when the condition is satisfied.

researchmap

Event date： 2012

Language：Japanese  

This paper presents the design of an anthropomorphic artificial finger. The robot finger features an underactuated mechanism which replicates three muscles and its tendons in human: flexor digitorum profundus, extensor digitorum, lumbrical. We analyze the motion of the robot finger from a quasi-static perspective. The simulated motion shows it is easy to change the motion by adjusting only one mechanical parameter. It achieves the preshaping with a self-adaptive robot finger. Finally, a prototype of a robot finger is built and tested to compare its motion with the simulated one.

researchmap

Event date： 2010

Language：Japanese  

The human can handle objects dexterously by feeding back the signals from tactile receptors. We made a prototype robot hand that has an anthropomorphic elastic skin which contains strain gauges and PVDF films as tactile receptors. This robot hand can detect slip between fingers and the grasped object using PVDF film information. In this paper, we make this robot hand learn outputs of own strain gauges when a slip occured by Neural Network. After learning, robot hand can adapt the grasp force to increasing load force of the grasped object and can prevent slip.

researchmap

Event date： 2010

Language：Japanese  

This paper describes learning of robust haptic recognition by Bionic Hand, a human-like robot hand, through dynamic interaction with the object. Bionic hand is a human-like hand that has soft skin with distributed receptors and is driven by artificial pneumatic muscles. At the beginning of learning, it utilizes the result of physical interaction with the object: thanks to the hand compliance, regrasping will leads object's posture to stable one in the hand. This result can be successively used as object classification for learning dynamic interaction beetween the hand and the object by a recurrent neural network. We conduct experiments and show that the proposed method is effective for robust and fast object recognition.

researchmap