小方 博之

Saffron pistils are normally harvested by hand. As the amount of pistils obtained from one flower is very small, harvesting them a certain amount requires a hard task. In this paper, we discuss a system to automate harvesting saffron pistil using robot equipped with image processing unit. We introduce a mechanism and a task movement of the robot, and examine technique to detect cutting point and gripping point which are necessary for harvesting by image processing, and proposes a process to harvest based on it.

We propose an algorithm to detect characters in images. The proposed algorithm consists of three phases: training, candidate area detection, and character detection. In training phase, a classifier to detect the character is generated from several haar-like feature-based weak classifiers by using Adaboost. In candidate area detection phase, one haar-like feature is used to define the area to search characters. And in character detection phase, the generated Adaboost-based classifier is used to detect the character in the defined area. We performed preliminary experiments to evaluate the proposed algorithm using 13 images. The experimental results show that some character on homogeneous background can be detected by this algorithm.

In this paper, we report an implementation of self-localization system for mobile robots running in outdoor environments where GPS data is sometimes not accurate enough. Our system is equipped with GPS, encoder and magnetic sensor. Data obtained by these sensors are integrated using Kalman filter. GPS observation is omitted when the robot is running through particular areas. These areas are determined based on standard deviation of the data observed in advance.

In late years tests called the CBT where I used a computer for by information technology and the network technology that developed spread. However, the knowledge examinations such as TOEFL or the Kanji Aptitude Test are often found, and these do not advance in the skill examination. Therefore, toward the CBT of the skill examination, I analyze the skill as the part. I set a problem for forward upward circling and, using an electromyograph, an acceleration sensor, the data of the camera, wrestle with the analysis of a necessary skill.

筆記をする際のペン持ち方に注目し,カメラで撮影したペン持ち方画像を用いて個人を認証する手法を検討する.ペンの持ち方は,手の大きさ等の人の身体的特徴に由来する個人性とともに,どのようにペンを持つのか,という人の癖に由来する個人性が存在すると考えられ,個人認証に有効なモダリティだと考えられる.本研究ではその有効性を確認するために,ペン持ち方を撮影した画像 (ペン持ち方データ) から多数の特徴を抽出し,各特徴から計算される非類似度を統合することで認証を行う.本研究では 33 個の特徴から計算される非類似度スコアを realAdaBoost を用いて統合し,ユーザ依存しきい値と比較することで認証を行った.30 人から取得したペン持ち方データを用いた評価実験では,他人の握り方を真似したなりすまし攻撃に対して等誤り率 4.1% という結果を得た.

人が筆記を行う際のペンの持ち方は個人ごとに異なっており, 個人認証に有効だと考えられる.本論文では, ペンの持ち方特徴を用いた個人認証手法を提案する.提案手法では, カメラを用いてペンの持ち方画像を撮影し, 複数の特徴をその画像から抽出する.抽出した特徴を事前登録されているデータと比較し非類似度ベクトルを計算し, それらを3層パーセプトロンで組み合わせることにより認証スコアを取得する.収集したデータベースを用いた実験ではなりすまし攻撃に対して等誤り率5.6%という結果になった.

Face detection algorithm used in Adaboost and Haar-like features were detected using the kana-kanji area detection. Create a set of images for learning. It was made by changing the number of sheets of the correct answer image and non-correct answer image. Correct answer images has Grayscale images and binary font images. Font used MS Gothic and MS Mincho. Non-correct answer images using gray-scale images of characters not included. The detection algorithm was verified by using 15 images for the verification. Verification images used 11 images including characters, two images including correct answer images, and two images not including characters. Adaboost and Haar-like feature using kana-kanji detection algorithms were validated.

Saffron pistils have been used, for example, as spice, dye and medicine from old days. But saffron pistils are considerably expensive, and are regarded to be luxury food. Saffron pistil is normally harvested by hand now. However, as the amount of pistils of one flower is very small, task of harvesting them is very hard. Therefore, in this paper, we discuss a method to automate harvesting saffron pistil using robot. We examine technique to detect cutting point and gripping point which are necessary for harvesting by online image processing. We verified the feasibility of our method through experiments.

家庭用掃除ロボットが効率的に掃除するための掃引アルゴリズムの研究を行う。掃除ロボットは部屋形状や家具などの条件により、狭い空間から抜け出せず、同じ場所を重複して掃引し、作業効率がしばしば低下する。本研究では、空間を広さに応じて2種類に分け、強化学習によってロボットのセンサ情報からどちらの空間にいるか判断し、その状態に適した走行方法を選択することで、掃引ムラを軽減し、掃引作業の効率化を目的とする。

当研究室では、家庭用掃除ロボットが効率的に掃除する為の掃引アルゴリズムの研究を行っている。その結果、部屋形状や家具などの配置により、狭い空間が形成されるとき、その部分を重複して掃引するため効率が落ちることがわかった。本研究ではゴミセンサを掃除用ロボットに搭載して狭所空間での重複掃引を検出し、作業のムラを無くすことを目的としています。

All stigmas of saffrons which are used as a spice or a dye are harvested by hand. However, it is heavy work to gather an enough amount. The purpose of this paper is to examine whether harvesting saffron's stigmas automatically is possible by using image processing. As a process of harvesting saffron stigmas automatically, we conceieved an idea of cutting the root of flower and harvesting stigma which is hanging by turning it down. Features to harvest stigmas are detected by using color information. The feasibility of this method was shown by the experiment.

Recently, cleaning robots for domestic use are developed. However, present algorithm cannot always sweep the workspace uniformly, and this causes the inefficiency of the algorithm. The best sweeping algorithm may depend on the shape of the room. To select the best algorithm, we examined a method to acquire the shape of a room in the last report. In this report, we adopt an actual robot model so that the robot can distinguish rooms in the real world.

Recently, cleaning robots for domestic use have been actively developed. In our past studies, we introduced a new sweeping algorithm for cleaning robots, and verified its efficiency on a simulator. In this report, we mount the algorithm on a real robot, to compare the performance with the simulation result.

Nowadays, home-use robots that clean floors automatically are developed. However, the efficiency of these robots is not so good. That is because the best way of cleaning room depends on the room's shape. If the robot knows which room it is cleaning now, it will be possible to make a robot cleaning the room more efficiently. Therefore we propose a method which the robot goes around the room with hug motion, to detect the outline of the room.

We propose a technique that automatically estimates a sports skill of examinees. In our study, we are discussing the technique for golf putting. However, we have estimated the skill only from apart of the examinee's motion data. It became the problem that the arbitrariness characteristics occur when choosing sensor. In this study, we introduce ensemble leaning to treat the whole motion data with Recurrent Neural Networks as temporary estimate sports skill and unify these result. The experimental result suggested the effectiveness of our method.

The form of the putt swing is different among person. This suggests that an advice that can effectively improve a person's skill may differ between them. This paper describes a method to classify putt swing form aiming to develop an advisory system. We obtained motion data of subjects using a motion capture system. The putt swing form was appropriately classified using cluster analysis.

移動ロボットに搭載されたカメラにより獲得される動画像から物体全体の形状を三次元復元するためには、任意の一視点方向からの復元だけでは不十分であり、移動経路において適宜最適な移動方向を選択しつつ、局所ごとの形状復元を行い、最終的に複数の局所的な復元結果を統合する必要がある。本報告では、移動カメラ画像からの三次元復元法として、モンテカルロ法とベイズ推定を用いる手法を用いて対象物体の局所的な三次元形状を復元する方法を検討する。更に、三次元形状復元を行った視点(動画像の初期フレームに対応)からの対象物体の見え方に関する情報量の取得を行い、ロボットの移動方向の選択をするための評価尺度としての有用性を検討する。

移動ロボットに搭載されたカメラにより獲得される動画像から物体全体の形状を三次元復元するためには、任意の一視点方向からの復元だけでは不十分であり、移動経路において適宜最適な移動方向を選択しつつ、局所ごとの形状復元を行い、最終的に複数の局所的な復元結果を統合する必要がある。本報告では、移動カメラ画像からの三次元復元法として、モンテカルロ法とベイズ推定を用いる手法を用いて対象物体の局所的な三次元形状を復元する方法を検討する。更に、三次元形状復元を行った視点(動画像の初期フレームに対応)からの対象物体の見え方に関する情報量の取得を行い、ロボットの移動方向の選択をするための評価尺度としての有用性を検討する。

移動ロボットに搭載されたカメラにより獲得される動画像から物体全体の形状を三次元復元するためには、任意の一視点方向からの復元だけでは不十分であり、移動経路において適宜最適な移動方向を選択しつつ、局所ごとの形状復元を行い、最終的に複数の局所的な復元結果を統合する必要がある。本報告では、移動カメラ画像からの三次元復元法として、モンテカルロ法とベイズ推定を用いる手法を用いて対象物体の局所的な三次元形状を復元する方法を検討する。更に、三次元形状復元を行った視点(動画像の初期フレームに対応)からの対象物体の見え方に関する情報量の取得を行い、ロボットの移動方向の選択をするための評価尺度としての有用性を検討する。

Recently, computer-based testing that automates examination using computer is attracting a lot of attention due to the development of information technology and of the spread of computer network. This paper describes a method to automate a test in which examinees actually perform the task to solve given problems, taking putt swing for example. Our testing system assesses the performance of examinees from motion data. We got motion data using a motion capture system. In this paper, we propose a method using recurrent neural network to estimate the expertise of examinees.

Recently, it is noticed that the Computer-Based Testing is getting a lot of attention as personal computers become popular and the internet society is growing. At present, CBT is normally used to measure examinees' knowledge or intelligence. But, it is not considered enough to measure examinees' skill for now. In this paper, we propose a method to measure examinees' skill taking putt swing for example. The task is easy to analyze and compare because the initial condition of the task can be fixed easily. We got motion data using a motion capture system. And we analyzed the motion data.

An irregular cleaning is caused easily because the home cleaning robot reflects information acquired while directly in operation. It is thought that the sweep rate prediction is effective to an irregularity of cleaning that is one of the problems of the home cleaning robot. However, whether it predictions is not verified in the environment that arranges furniture. In this research, it is verified whether the prediction of sweep rate is possible in the environment that arranges furniture, and the influence of furniture in the sweep rate prediction of the cleaning robot is described.

Recently, robot cleaner used at home is developing popular. But performance of these robots on sale at present isn't enough, because sweeping by these robots make problems such as irregularities of a sweep. We think that an inefficiency of a sweep in narrow environment influences it as those causes. In the report, we analyzed sweep work in narrow environment and examined influence to a sweep.

It is important for robots to recognize unregistered objects when working in general environments like home or office. In such a case, the information of shape and material may play an important role to recognize them. In this research, a sequential method of inferring object's surface material using brightness data from a series of images is examined.

The probabilistic road map method in one of the good methods to generate trajectories of manipulators with multiple degrees of freedom. However, there is a problem that the subgoal cannot be put well, hi this research to solve such a problem, the Markov chain Monte Carlo method is applied to make the probabilistic road map method more efficient, and the effectiveness of the proposed method is confirmed by simulation.

Recently, the automation of the test using computer is emerging because of the development of information technology. This paper describes the method to automate tests in which examinees actually perform the task to solve given problems. Our testing system recognizes performance of the examinee with sensors such as cameras. The uncertainty of recognition is a non-negligible problem, here.hi this paper, we discuss a method to estimate the expertise of examinees considering such an uncertainty. The validity of this method is verified by using simulation.

今後,整備されていない環境下でロボットが作業する機会が増加すると考えられる.そのような環境でロボットが作業する場合,未知の物体が作業対象となり,それらを識別しなければならなくなる.その際,形状や材質などが,物体を識別する上で重要な情報となる.前報では我々はロボットのカメラから得られる連続画像を用いて物体形状を復元する方法について検討を行った.本報では,連続画像の輝度データから物体表面材質を反射モデルを利用して推定する方法を述べる.

Nowadays, many mobile robots are developed to automate cleaning task. Some of them aim at using at home. Most of the home cleaning robots are using reactive type algorithm. However, when a robot adopting reactive type algorithm gets into a narrow area surrounded by obstacles, the robot cannot readily get out of there. This causes the inefficiency of the cleaning task. We propose efficient hybrid algorithm to solve this problem.