In this paper, we propose a power assist suit for supporting and transporting a heavy object with a single motor that reduces the load on the whole body of a wearer. The effectiveness of the power assist suit was verified by experiments.

Conventional methods of gait analysis for person identification use features extracted from a sequence of camera images taken during one or more gait cycles. The walking direction is implicitly assumed not to change. However, with the exception of very particular cases, such as walking on a circle centered on the camera, or along a line passing through the camera, there is always some degree of orientation change, most pronounced when the person is closer to the camera. This change in the angle between the velocity vector and the position vector in respect to the camera causes a decrease in performance for conventional methods. To address this issue we propose in this paper a new method, which provides improved identification in this context of orientation change. The proposed method uses a 4D gait database consisting of multiple 3D shape models of walking people and adaptive virtual image synthesis with high accuracy. Each frame, for the duration of a gait cycle, is used to estimate the walking direction of the subject, and a virtual image corresponding to the estimated direction is synthesized from the 4D gait database. The identification uses affine moment invariants as gait features. The efficiency of the proposed method is demonstrated through experiments using a database that includes 42 subjects. (C) 2014 Elsevier B. V. All rights reserved.

We propose a novel method for recognizing sequential patterns such as motion trajectory of biological objects (i.e., cells, organelle, protein molecules, etc.), human behavior motion, and meteorological data. In the proposed method, a local classifier is prepared for every point (or timing or frame) and then the whole pattern is recognized by majority voting of the recognition results of the local classifiers. The voting strategy has a strong benefit that even if an input pattern has a very large deviation from a prototype locally at several points, they do not severely influence the recognition result; they are treated just as several incorrect votes and thus will be neglected successfully through the majority voting. For regularizing the recognition result, we introduce partial-dependency to local classifiers. An important point is that this dependency is introduced to not only local classifiers at neighboring point pairs but also to those at distant point pairs. Although, the dependency makes the problem non-Markovian (i.e., higher-order Markovian), it can still be solved efficiently by using a graph cut algorithm with polynomial-order computations. The experimental results revealed that the proposed method can achieve better recognition accuracy while utilizing the above characteristics of the proposed method.

Conventional methods of gait analysis for person identification use features extracted from a sequence of camera images taken during one or more gait cycles. An implicit assumption is made that the walking direction does not change. However, cameras deployed in real-world environments (and often placed at corners) capture images of humans who walk on paths that, for a variety of reasons, such as turning corners or avoiding obstacles, are not straight but curved. This change of the direction of the velocity vector causes a decrease in performance for conventional methods. In this paper we address this aspect, and propose a method that offers improved identification results for people walking on curved trajectories. The large diversity of curved trajectories makes the collection of complete real world data infeasible. The proposed method utilizes a 4D gait database consisting of multiple 3D shape models of walking subjects and adaptive virtual image synthesis. Each frame, for the duration of a gait cycle, is used to estimate a walking direction for the subject, and consequently a virtual image corresponding to this estimated direction is synthesized from the 4D gait database. The identification uses affine moment invariants as gait features. Experiments using the 4D gait database of 21 subjects show that the proposed method has a higher recognition performance than conventional methods.

Facial animation is one of the most important contents in 3D CG animations. By the development of scanning and tracking methods of a facial motion, a face model which consists of more than 100,000 points can be used for the animations. To edit the facial animations, key point based approaches such as "face rigging" are still useful ways. Even if a facial tracking method gives us all point-to-point correspondences, a detection method of a suitable set of key points is needed for content creators. Then, we propose a method to detect the key points which efficiently represent motions of a face. We optimize the key points for a Radial Basis Function (RBF) based 3D deformation technique. The RBF based deformation is a common technique to represent a movement of 3D objects in CG animations. Since the key point based approaches usually deform objects by interpolating movements of the key points, these approaches cause errors between the deformed shapes and the original ones. To minimize the errors, we propose a method which automatically inserts additional key points by detecting the area where the error is larger than the surrounding area. Finally, by utilizing the suitable set of key points, the proposed method creates a motion of a face which are transferred form another motion of a face.

This paper proposes a new dynamic time warping (DTW) method, called non-Markovian DTW In the conventional DTW, the warping function is optimized generally by dynamic programming (DP) subject to some Markovian constraints which restrict the relationship between neighboring time points. In contrast, the non-Markovian DTW can introduce non-Markovian constraints for dealing with the relationship between points with a large time interval. This new and promising ability of DTW is realized by using graph cut as the optimizer of the warping function instead of DP. Specifically, the conventional DTW problem is first converted as an equivalent minimum cut problem on a graph and then edges representing the non-Markovian constraints are added to the graph. An experiment on online character recognition showed the advantage of using non-Markovian constraints during DTW.

本論文では,ノードの出力メッセージをそのノードの全出力メッセージの平均値で置き換え,更に低解像度グラフから元のグラフへと階層的にメッセージを伝搬することにより,画像に対して適用した場合に通常の確率伝搬法と比較して計算時間を約2分の1から3分の1に,メモリ使用量を40%に削減可能な近似確率伝搬法を提案する.また,提案手法をCPU及びGPU上に実装し,Middleburyステレオデータセットを使用して通常の確率伝搬法と比較実験を行い,わずかな精度の低下と引換えに計算時間とメモリ使用量の両面において性能が大きく改善されたことを示す.

Frequent patterns in time series data are useful clues to learn previously unknown events in an unsupervised way. In this paper, we propose a method for detecting frequent patterns in long time series data efficiently. The major contribution of the paper is two-fold: (1) Partly Locality Sensitive Hashing (PLSH) is proposed to find frequent patterns efficiently and (2) the problem of finding consecutive time frames that have a large number of frequent patterns is formulated as a combinatorial optimization problem which is solved via Dynamic Programming (DP) in polynomial time <i>O</i> (<i>N</i> ^1+1/&alpha;) thanks to PLSH where <i>N</i> is the total amount of data. The proposed method was evaluated by detecting frequent whole body motions in a video sequence as well as by detecting frequent everyday manipulation tasks in motion capture data.

Repeated patterns are useful clues to learn previously unknown events in an unsupervised way. This paper presents a novel method that detects relatively long variable-length unknown repeated patterns in a motion sequence efficiently.
The major contribution of the paper is two-fold: (1) Partly Locality Sensitive Hashing (PLSH) [1] is employed to find repeated patterns efficiently and (2) the problem of finding consecutive time frames that have a large number of repeated patterns is formulated as a combinatorial optimization problem which is solved via Dynamic Programming (DP) in polynomial time O( N1+1/alpha) thanks to PLSH where N is the total amount of data. The proposed method was evaluated by detecting repeated interactions between objects in everyday manipulation tasks and outperformed previous methods in terms of accuracy or computational time.

This paper presents an approximate belief propagation algorithm that replaces outgoing messages from a node with the averaged outgoing message and propagates messages from a low resolution graph to the original graph hierarchically. The proposed method reduces the computational time by half or two-thirds and reduces the required amount of memory by 60% compared with the standard belief propagation algorithm when applied to an image. The proposed method was implemented on CPU and GPU, and was evaluated against Middlebury stereo benchmark dataset in comparison with the standard belief propagation algorithm. It is shown that the proposed method outperforms the other in terms of both the computational time and the required amount of memory with minor loss of accuracy. © 2010 IEEE.

In this paper, we propose a method that detects repeated motion patterns in a long motion sequence efficiently. Repeated motion patterns are the structured information that can be obtained without knowledge of the context of motions. They can be used as a seed to find causal relationships between motions or to obtain contextual information of human activity, which is useful for intelligent systems that support human activity in everyday environment.
The major contribution of the proposed method is two-fold:
(1) motion density is proposed as a repeatability measure and
(2) the problem of finding consecutive time frames with large motion density is formulated as a combinatorial optimization problem which is solved via Dynamic Programming (DP) in polynomial time O(N log N) where N is the total amount of data. The proposed method was evaluated by detecting repeated interactions between objects in everyday manipulation tasks and outperformed the previous method in terms of both detectability and computational time.

This paper presents a supervised foreground segmentation method that uses local and global feature similarity with edge constraint. This framework integrates and extends the notion of region growing and classification to deal with local and global fitness. It parameterizes constraint of growing using Chebyshev's inequality. The constraint is used to stop segmentation before matting. Matting relies on both local and global information. The proposed method outperforms many of the current methods in the sense of correctness and minimal user interaction, and it does so in a reasonable computation time.

In this paper, we propose a method for estimating meaningful actions from long-term observation of everyday manipulation tasks without prior knowledge as part of an action understanding framework for life support robotic systems. The target task is defined as a sequence of interactions between objects. An interaction that appears many times is assumed to be meaningful and repetitious relative motion patterns are detected from trajectories of multiple objects. The main contribution is that the problem is formulated as a combinatorial optimization problem with two parameters, target object labels and correspondences on similar motion patterns, and is solved using local and global Dynamic Programming (DP) in polynomial time O(N log N), where N is a total amount of data. The proposed method is evaluated against manipulation tasks using everyday objects such as a cup and a tea-pot.

This paper describes a painting robot with multi-fingered hands and stereo vision. The goal of this study is for the robot to reproduce the whole procedure involved in human painting. A painting action is divided into three phases: obtaining a 3D model, composing a picture model, and painting by a robot. In this system, various feedback techniques including computer vision and force sensors are used. As experiments, an apple and a human silhouette are painted on a canvas using this system.

This paper presents a supervised foreground segmentation method that uses local and global feature similarity with edge constraint. This framework integrates and extends the notion of region growing and classification to deal with local anti global fitness. It parameterizes constraint of growing using Chebyshev's inequality. The constraint is used to stop segmentation before matting. Matting relies on both local and global information. The proposed method outperform many of the current methods in the sense of correctness and minimal user interaction, and it does so in a reasonable computation time.

This paper presents a novel whole body motion estimation method by fitting a deformable articulated model of the human body into the 31) reconstructed volume obtained from multiple video streams. The advantage of the proposed method is two fold: (1) combination of a robust estimator and ICP algorithm with Kd-tree search in pose and normal space make it possible to track complex and dynamic motion robustly against noise and interference between limb and torso, (2) the hierarchical estimation and backtrack re-estimation algorithm enable accurate estimation.
The power to track challenging whole body motion in real environment is also presented.

This paper presents visual perception discovered in high-level manipulator planning for a robot to reproduce the procedure involved in human painting. First, we propose a technique of 3D object segmentation that can work well even when the precision of the cameras is inadequate. Second, we apply a simple yet powerful fast color perception model that shows similarity to human perception. The method outperforms many existing interactive color perception algorithms. Third, we generate global orientation map perception using a radial basis function. Finally, we use the derived foreground, color segments, and orientation map to produce a visual feedback drawing. Our main contributions are 3D object segmentation and color perception schemes.

This paper presents visual perception discovered in high-level manipulator planning for a robot to reproduce the procedure involved in human painting. First, we apply a technique of 2D object segmentation that considers region similarity as an objective function and edge as a constraint with artificial intelligent used as a criterion function. The system can segment images more effectively than most of existing methods, even if the foreground is very similar to the background. Second, we propose a novel color perception model that shows similarity to human perception. The method outperforms many existing color reduction algorithms. Third, we propose a novel global orientation map perception using a radial basis function. Finally, we use the derived model along with the brush's position- and force-sensing to produce a visual feedback drawing. Experiments show that our system can generate good paintings including portraits.

In this paper, we describe a painting robot with multi-fingered hands and stereo vision. The goal of this study is for the robot to reproduce the whole procedure involved in human painting. A painting action is divided into three phases: obtaining a 3D model, composing a picture model, and painting by a robot. In this system, various feedback techniques including computer vision and force sensors are used. As experiments, an apple and a human silhouette are painted on a canvas using this system. © 2006 IEEE.

レーザレンジセンサなどを用いた計測作業において, 遺跡や建物などの対象物が大規模になればなるほど, 地上からは観測できないオクルージョン領域が発生する. 一般的には足場を組んでセンサを持ち上げてその上部からの計測を行っているが, 計測位置の移動には多くの時間とコストがかかる. また, 足場を組むことが不可能な場合は計測作業自体が止まることにもなる. この問題を解決するために, 筆者らは気球を用いて空中から計測を行うことにより, 足場を組むことなく手軽に計測位置の移動を可能にするレーザレンジセンサFLRSの開発を行った. このセンサでは, 気球の揺れによる影響を少なくするために高速な計測を実現しているが, 計測結果には若干のひずみが発生している. そこで, センサに内蔵しているビデオカメラからの連続画像を用いて, センサ本体の動きを検出して補正する手法を提案する. そして, 実際に遺跡計測に用いた例を挙げ, センサシステムの有効性を示す.

The<I>Learning from Observation</I> (LFO) paradigm has been widely applied in various types of robot systems. It helps reduce the work of the programmer. However, so far, the available systems have application limited to rigid objects. Deformable objects are not considered, because: (1) it is difficult to describe their states and (2) too many operations are possible on them. In this paper, we choose knot-tying tasks as case study for operating on deformable objects, since the<I>knot theory</I>is available and since types of operations are limited. In actuality, we introduce an appropriate representation for describing knot's states and define four types of operations enough to realize any knot-tying tasks.

The Programming by Demonstration (PbD) technique aims at teaching a robot to accomplish a task by learning from a human demonstration. In a manipulation context, recognizing the demonstrator's hand gestures, specifically when and how objects are grasped, plays a significant role. Here, a system is presented that uses both hand shape and contact-point information obtained from a data glove and tactile sensors to recognize continuous human-grasp sequences. The sensor fusion, grasp classification, and task segmentation are made by a hidden Markov model recognizer. Twelve different grasp types from a general, task-independent taxonomy are recognized. An accuracy of up to 95% could be achieved for a multiple-user system. © 2005 IEEE.

In this paper, we propose an efficient and effective image generation system for "Mixed Reality Traffic Experiment Space", an enhanced driving/traffic simulation system which we have been developing for Sustainable ITS project at the University of Tokyo. Conventional driving simulators represent ther view by a set of polygon-based objects, which leads to less photo-reality and huge human costs for dataset construction. We introduce our image/geometry-based hybrid method to realize more photo-realistic view with less human cost at the same time. Images for datesets are captured from real world by multiple video cameras mounted on a data acquisition vehicle. And the view for the system is created by synthesizing the image dataset. Following contents mainly describe details on data acquisition and view rendering. © 2005 IEEE.

We propose an efficient and effective image generation system for an experimental mixed-reality traffice space. Our enhanced traffic/driving simulation system represents the view through a hybrid that combines virtual geometry with real images to realize high photo-reality with little human cost. Images for datasets are captured from the real world, and the view for the simulation system is created by synthesizing image datasets with a conventional driving simulator. © 2005 IEEE.

The <I>planning-from-observation</I> paradigm is widely noticed as a novel robot-programming technique. It consists of two parts: (1) recognition of a human demonstration from observation as symbolic representation, i.e., a sequence of movement primitives and (2) execution of the same task. Symbolic representation enables a robot to achieve the same task even in a different environment. We already proposed a method to build a symbolic representation for an assembly task. However, for a robot to execute the task, it is necessary to adjust parameters of each movement primitive to generate appropriate trajectories of objects, especially a path which maintains one of any contact states. Many researchers have proposed methods to calculate such a path using a non-linear optimization method, for example, the potential field method, the probabilistic method, and so on. Although these methods are very powerful as calculation tools on a computer, their solutions are not optimal. We propose a novel method to calculate such an optimal path using a linear solution.

Learning from Observation (LFO) has been widely applied in various types of robot system. It helps reduce the work of the programmer. But the available systems have application limited to rigid objects. Deformable objects are not considered because: 1) it is difficult to describe their state and 2) too many operations are possible on them. In this paper, we choose the knot tying as case study for operating on nonrigid bodies, because a "knot theory" is available and the type of operations is limited. We describe the Knot Planning from Observation (KPO) paradigm, a KPO theory and a KPO system.

As a way of automatic programming of robot behavior, a method for building a symbolic manipulation task model from a demonstration is proposed. The feature of this model is that it explicitly stores the information about the essential parts of a task, i.e. interaction between a hand and an environmental object, or interaction between a grasped object and a target object. Thus, even in different environments, this method reproduces robot motion as similar as possible to that of humans to complete the task while changing the motion during non-essential parts to adapt to the current environment. To automatically determine the essential parts, a method called attention point analysis is proposed
this method searches for the nature of a task using multiple sensors and estimates the parameters to represent the task. A humanoid robot is used to verify the reproduced robot motion based on the generated task model.

This paper describes a new approach on how to teach everyday manipulation tasks to a robot under the "Learning from Observation" framework. In our previous work, to acquire low-level action primitives of a task automatically, we proposed a technique to estimate essential interactions to complete a task by integrating multiple observations of similar demonstrations. But after many demonstrations are performed, there are possibly interactions which are the same in nature. These identical interactions should be grouped so that each action primitive becomes unique. For this purpose, a Hidden Markov Model based clustering algorithm is presented which automatically determines the number of the independent interactions, We also show that obtained interactions can be used as discriminations of human behavior. Finally, a simulation result and an experimental result in which a real humanoid robot learns and recognizes essential actions by observing demonstrations are presented.

The completion of robot programs requires long development time and much effort. To shorten the programming time and to minimize the effort, we have been developing a system which we refer to as the "assembly-plan-from-observation (APO) system." This system requires assembly task recognition from observing human performance. Observation data by a robot's vision system is usually error contaminated, and, thus, we cannot use those data directly. This paper proposes two methods to clean up those errors by using contact relations and their transitions. The first one corrects the observed configuration from contact relations observed. The second one identifies wrongly determined contact relations from an analysis of configuration space (C-space). We have implemented both methods on our test bed and have verified their effectiveness.

This paper describes our current research on learning task level representations by a robot through observation of human demonstrations. Representations proposed so far typically utilized one or a few sensors and were constructed through one time observation, therefore they could solve ambiguity in the data only under strong assumptions. We focus on human hand actions and have been developing a novel construction method of a human task model which integrates multiple observations to solve ambiguity based on Attention Points (APs). So far, this analysis constructs a symbolic task model efficiently in a coarse-to-fine way through two steps. However, to represent delicate motion appeared in a task, the system must incorporate the information about precise motion of the manipulated objects into an abstract task model. In this paper, we propose a method to identify the manipulated object through repeated observations of both human and robot behavior. To this end, we present a method which combines 2D and 3D template matching techniques to localize an object in 3D space generated from a depth and an intensity image. We apply this technique to recognition of human and robot behavior by obtaining the precise trajectory of the manipulated objects. We also present the experimental results achieved through the use of a human-form robot equipped with a 9-eye stereo vision system.

This paper describes our current research on learning task level representations by a robot through observation of human demonstrations. We focus on human hand actions and represent such hand actions in symbolic task models. We propose a framework of such models by efficiently integrating multiple observations based on attention points
we then evaluate the produced model by using a human-form robot. We propose a two-step observation mechanism. At the first step, the system roughly observes the entire sequence of the human demonstration, builds a rough task model and also extracts attention points (APs). The attention points indicate the time and the position in the observation sequence that requires further detailed analysis. At the second step, the system closely examines the sequence around the APs, and obtains attribute values for the task model, such as what to grasp, which hand to be used, or what is the precise trajectory of the manipulated object. We have implemented this system on a human form r obot and demonstrated its effectiveness.

This paper describes our current research on how a robot, through observation of human demonstrations, can learn task level representations of human hand-work tasks. Representations proposed so far typically handle a human hand trajectory as it is or segment the entire task into a discrete pre-determined symbol sequence
but to make a generalized model of human hand-work tasks, both types of information must be incorporated into the model appropriately. We propose a technique for segmenting an observed hand-work task into pieces which are composed of fine motion or coarse motion. Fine motion means delicate manipulation and holds the relative trajectory between the grasped object and the target object, while coarse motion is a symbol which connects each fine motion. During coarse motion, a trajectory can be adjusted according to the environment or the structure of a robot when the robot performs the same task performed by a human. To extract essential fine motion automatically, we propose a technique for integrating and aligning multiple observations of different demonstrations, which are virtually the same task, by using data gloves and multi-dimensional Dynamic Programming (DP) matching. Along each fine motion, the relative trajectory (position and orientation) is calculated by tracking the manipulated object using a stereo vision. To localize and track the manipulated object efficiently, we propose a model-based localization technique, which combines 2D and 3D template matching. We have implemented those techniques on our human-form robot and present an experimental result which analyzed and performed a non-contact hand-work task. © 2001 IEEE.

This paper presents a novel method of constructing a human behaviour model by attention point (AP) analysis. The AP analysis consists of two steps. At the first step, it broadly observes human behaviour, constructs rough human behaviour model and finds APs which require detailed analysis. Then at the second step, by applying time-consuming analysis on APs in the same human behaviour, it can enhance the human behaviour model. This human behaviour model is highly abstracted and is able to change the degree of abstraction adapting to the environment so as to be applicable in a different environment. We describe this method and its implementation using data gloves and a stereo vision system. We also show an experimental result in which a real robot observed and performed the same human behaviour successfully in a different environment using this model.

This paper presents a novel method of constructing a human task model by attention point (AP) analysis. The AP analysis consists of two steps. At the first step, it broadly observes human task, constructs rough human task model and finds APs which require detailed analysis. Then at the second step, by applying time-consuming analysis on APs in the same human task, it can enhance the human task model. This human task model is highly abstracted and is able to change the degree of abstraction adapting to the environment so as to be applicable in a different environment. We describe this method and its implementation using data gloves and a stereo vision system. We also show an experimental result in which a real robot observed a human task and performed the same human task successfully in a different environment using this model.

The L-shaped gripper commonly used in power assist suits has the problem of not being able to stably hold arbitrarily shaped objects. To solve this problem, we developed a gripper with a wire-driven ratchet mechanism. However, we could not confirm its effectiveness in reducing the burden on the hands and arms for irregularly shaped objects such as rice bags. This is because the mechanism using a wire did not work as expected due to friction between the wire and the wire guide. Therefore, in this study, we solved these problems by developing a gripper with a link-driven ratchet mechanism, which does not use a wire. We conducted experiments to compare the proposed gripper with our previous gripper by measuring the surface EMG of the wearer's arms when lifting objects of various shapes.

The goal-conditioned reinforcement learning approach is useful when an agent wants to adaptively achieve arbitrary goals in various environments. Furthermore, it is desirable to be able to learn the behavior of an agent from observations from on-board sensors such as RGB cameras. Since it is difficult for agents to learn directly from images, previous research has shown that the learning process can be accelerated by learning latent representations using VAEs and performing goal-conditioned reinforcement learning in the latent space. However, latent representations learned using VAEs with image reconstruction loss contain information irrelevant to the task to be accomplished. In goal-conditioned reinforcement learning, the reward function is defined as the distance to the goal state, so latent representations that capture the distance between states are appropriate. In this study, we propose a method for maintaining the similarity between states in the latent space using contrastive learning and performing goal-conditioned reinforcement learning in the latent space given image observations. We compared the proposed method with that using latent representations obtained from VAEs, and show that our method outperformed the other.

In this paper, we present a mechanical design of a winch-type power assist suit that realizes luggage lifting movement and knee fixation-and-release movement with a single motor. To solve the problems of our previous power assist suit, we realized the descent of a load of 20kg, reduced the weight, improved the safety, and strengthened the rigidity. We developed a prototype power assist suit and verified if it works as designed. We also conducted lifting experiments and confirmed that it could lift a load of 5kg. We measured the surface electromyogram (EMG) and confirmed that it reduced the burden on the wearer's arms and back when lifting a load.

<p>L-shaped grippers typically used for power assist suits have a problem in stably holding objects of arbitrary shape. To solve this problem, we developed a gripper with a ratchet mechanism consisting of linear motion and rotational motion parts. The gripper can be deformed according to the shape of the wearer's hand and the shape of the object to be held, and the shape of the gripper can be fixed due to the rachet mechanism, thus objects of arbitrary shape can be held stably by this gripper and the burden of the wearers is reduced. We carried out experiments where four types of objects were held by the developed gripper. We showed that the developed gripper has reduced the wearer's burden for two types of objects among four.</p>

<p>In this paper, we present a power assist suit that supports and transports heavy objects with a single motor. The leg mechanism of the power assist suit is made of upper and lower parallel links and springs. It is designed to be located inside the wearer's legs so that the power assist suit becomes small and light-weight, and the stability during transportation is improved. We measured the surface electromyogram (EMG) during luggage lifting experiments and confirmed that the burden of the wearer's back and legs is reduced when wearing the developed power assist suit.</p>

<p>In this paper, we present a mechanical design of a wire drive mechanism for power assist suits that realizes linear luggage lifting movement and nonlinear knee fixation-and-release movement using a motor. The nonlinear fixation-and-release movement is realized by a mechanism using a piston and a crankshaft. We developed an algorithm to find the design parameters that minimize the error from the optimal wire movement. We developed a prototype wire drive mechanism and verified that it works as designed. In addition, we carried out lifting experiments and confirmed that luggage up to 20 kg can be lifted. We measured the surface electromyogram (EMG) and confirmed that the burden of the wearer's arms was reduced during luggage lifting.</p>

In this paper, we present a mechanical design of an exoskeleton-type power assist suit that helps a wearer to lift up and transport heavy objects while walking. The assist suit supports heavy objects by itself thus the force exerting on the whole body of the wearer is reduced. In addition, the hip and leg joint axes are aligned with those of the wearer so that the wearer can walk smoothly. The experimental results show that the developed power assist suit is able to support and transport heavy objects while walking without exerting a load on the upper body of a wearer including the arms and legs.

歩行画像列を用いた個人識別手法では,例えば対象人物のカメラに対する歩行方向がデータベース構築時と異なる場合,対象人物の見えが変化するため識別率が低下する.これに対し我々は,対象人物の3次元モデル列から構成される4次元歩容データベースを構築し,歩行方向による見えの変化に頑強な個人識別手法を提案した.この手法ではまず,複数台のカメラから撮影された歩行画像から,視体積交差法により対象人物の3次元形状を復元する.次に複数の仮想的な任意視点から3次元モデルの仮想視点画像を合成し,各仮想視点において抽出された歩行特徴により学習データを構築した.個人識別時には,対象人物の歩行画像から歩行特徴を抽出し,これを学習データの全ての歩行特徴と比較することで,個人識別を行った.しかし,全ての仮想視点における歩行特徴を用いているため,計算コストが高く,また実際とは異なる仮想視点が選択される場合があった.そこで本論文では,まず撮影したカメラの視点をフリーズパターンにより推定し,その視点での歩行特徴のみを比較することで,計算コストの低下と識別性能の向上を実現する個人識別手法を提案する.実際に4次元歩容データベースを用いて個人識別実験を行い,提案手法の有効性を示す.

本稿では,時間的非線形伸縮を伴う時系列パターンの認識手法を提案する.本手法では,入力パターンの各サンプル点(各時刻)にクラスラベルを割り当てる.同時に,時間的非線形伸縮に対応するため,各サンプル点に標準パターン上において対応するサンプル点を割り当てる.入力パターンのクラスは,最終的に,クラスラベル数の多数決により確定される.割り当てに際し,入力パターンの時間的に隣接するサンプル点同士に相互制約を設け,できるだけ同じクラスにラベリングすると同時に,標準パターンのできるだけ時間的に隣接したサンプル点に対応付ける.本稿では,将来的な非隣接サンプル点間制約の導入をにらんで,グラフの最小切断(グラフカット)の一種であるα拡張アルゴリズムによりこの最適割り当て問題を解く.このアルゴリズムは多項式時間計算量で済む半面,原問題の近似解法となっている.そこで,求まる近似解がどの程度妥当かについて,オンライン文字データを対象とした認識実験を行い検証する.

本論文では，物体操作作業を２物体間の時系列相対運動パターンとみなし，これをあらかじめ用意した参照相対運動パターン集合と比較することによって，動画像から物体操作を認識する手法を提案する．ある時刻の画像に対し，画像中の任意の２領域がある参照相対運動パターンに属する尤度を，(1)１つ目の領域が１番目の物体である尤度，(2)２つ目の領域が２番目の物体である尤度，(3)領域間の相対位置，の３つの指標により定義する．本手法では，この尤度に基づき動的計画法を用いて入力動画像全体と各参照相対運動パターンとの比較を行い，最も類似した参照相対運動パターンが表す物体操作を認識結果として出力する．

査読

日本機械学会学術誌に投稿された論文の査読

査読

計測自動制御学会論文集に投稿された論文の査読

小・中・高校生を対象とした学部体験入学・出張講座等

近畿大学工業高等専門学校からインターンシップを１名受け入れ、ロボット機構の設計製作とマイコンによる制御に関する指導を行った。,日付:8月21日～8月25日

公開講座・講演会の企画・講師等

「ロボットアームと多指ハンドの動作計画」に関する講演,日付:3月16日

小・中・高校生を対象とした学部体験入学・出張講座等

大阪府大高専からインターンシップを１名受け入れ、ロボット機構の設計製作とマイコンによる制御に関する指導を行った。,日付:8月22日～8月26日

公開講座・講演会の企画・講師等

３次元計測技術のシーズに関する講演,日付:2015年11月10日

公開講座・講演会の企画・講師等

パワーアシストスーツに関するシーズの発表,日付:2015年11月9日

学術雑誌等の編集委員・査読・審査員等

Associate Editor,任期:2015.3-2018.2

学術雑誌等の編集委員・査読・審査員等

常任査読委員 ,任期:2010.8～

研究成果に係る新聞掲載、テレビ・ラジオ出演

研究紹介記事.
「教育・理科大好き!・物まねロボット」

学協会、政府、自治体等の公的委員

ロボティクス・メカトロニクス部門 広報委員会,任期:2015.4-2019.3

学協会、政府、自治体等の公的委員

PRMU研究専門委員会,任期:2014.5-2020.4

学協会、政府、自治体等の公的委員

研究運営委員会,任期:2008.4-2012.3

学術雑誌等の編集委員・査読・審査員等

委員