-
平山 泰寛, 辻内 伸好, 伊藤 彰人, 渥美 航
セッションID: 1P1-J02
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In order to enable the robot to automate the work of operating the string-shaped flexible object, we construct a model of the string-shaped flexible object and evaluate the accuracy of the model. In this study, a rope with a simple shape is used as the object of operation, and a multi-link model based on the dynamics of a rigid body is applied for modeling. Numerical analysis software MATLAB is used for model creation and simulation to express the dynamic shape of the target object. In this study, model parameters that behave in the same way as the work target are identified using the simplex method, which is an optimization method. In addition, the shape and model when the target object is deformed using a robot arm are compared, and the reproduction accuracy on the desk surface is evaluated.
抄録全体を表示
-
山本 千隼, 山脇 輔, 洞出 光洋, 八島 真人
セッションID: 1P1-J04
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper propose the learning scheme for a throwing manipulation on a 2-D workspace with the 1-DOF robotic arm and verifies its validity experimentally. The proposed learning scheme consists of Bayesian optimization and sequential quadratic programming. The proposed learning scheme is able to make the positioning error less than 0.7mm after 11 trials.
抄録全体を表示
-
巖 和貴, 小山 佳祐, 万 偉偉, 西 卓郎, 原田 研介
セッションID: 1P1-J05
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a method to automatically plan the motion of a robot that packs many kinds of objects into a box, assuming multiple grippers and considering whether the grippers can actually grasp and pack the objects into a box. An initial solution was created by using a GA like algorithm taking multiple grippers into account, and the optimal position of each object was determined by repeatedly improving the solution. Then, the packing motion planning was executed for each object. The effectiveness of the proposed method was confirmed by experiments.
抄録全体を表示
-
松谷 祐希
セッションID: 1P1-J06
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a new position control of a two-link arm using virtual musculoskeletal structure. The proposed method is able to accomplish a robust for the sensor information including a time delay.
抄録全体を表示
-
末村 萌, 笹竹 晴萌, 田崎 良佑
セッションID: 1P1-J07
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a method of tightening pressure estimation of multiplex string and binding robot manipulation considering the flexibility of the collective object. In order to enable applications to use for various purposes such as but not limited to space development and industrial production, out proposed algorithm which estimates the elastic state is designed by applying visual recognition and execute motion-planning. Specifically, the key phase of our research can be divided into three steps: 1) visual recognition, 2) flexibility estimation and motion-planning, 3) execution of binding motion. At the end, the goal is to develop a basic application methodology that enables an articulated robot to tie up flexible objects such as hair, using rubber strings.
抄録全体を表示
-
高須 瑞輝, 原田 研介, 万 偉偉, 小山 佳祐
セッションID: 1P1-J08
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
When we pick up one object in bulk, there is a possibility that tangles occur due to the shape of the object. In this research, we propose a method to select an object that has no or few possibilities of tangle and grasp it. We consider the case of multiple objects composed of a single material. In this study, we segment the hyperspectral images, convert them into fine lines and point clouds, and interpolate the 3D information. Based on these information, the GLI is used to quantitatively understand the entanglement of each object and the grasp position is determined using the Graspability. By applying this method, we confirmed that the robot can grasp one object in bulk which is difficult to be recognized by the conventional camera without tangling.
抄録全体を表示
-
長門 秀征, 元田 智大, 小山 佳祐, 万 偉偉, 原田 研介
セッションID: 1P1-J09
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a method for retrieving a target object by a robot in a complex environment. In such an environment, it becomes difficult to observe the state of the target object owing to occlusion, and to grasp it. In this work, we propose a method for selecting the viewpoint to observe the occlusion part of the target object by using RGB-D images and selecting the motion from grasping/dragging to extract the target object according to the situation. Finally, we confirmed that a robot can retrieve a certain target from the complex environments.
抄録全体を表示
-
Zhengtao Hu, Weiwei WAN, Keisuke KOYAMA, Kensuke HARADA
セッションID: 1P1-J10
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This work presents the manipulation policies for a robot to use a designed rotating tool to conduct screw fastening tasks. The tool converts the parallel motion of robotic grippers into continuous rotation, thus, reduces the difficulties of performing screwing motion. However, using the tool to complete a whole screw fastening task consisting of continuous screwing, firmly tightening, and detaching the tool from a screw head is a challenging work. To solve the problem, we formulate the screwing process as a decision problem and develop the policy-selection algorithm, which considers the actions of both the robotic gripper and robot arm. The experiments validate that with the proposed methods, a robot can use the tool to complete a screw fastening task successfully.
抄録全体を表示
-
藤田 剛輝, 平尾 和睦, 梅村 歩, Mochammad ARIYANTO, 森島 圭祐
セッションID: 1P1-J11
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
For the practical application of insect cyborg, which is a fusion of life and machine, we conducted research aimed at establishing the connection method between living body and machine and the law of biological control. In this study, we developed a jig to increase the connection yield and conducted a sensory organ stimulus response experiment to estimate a model of an insect cyborg. In the connection method using the manufactured jig, the connection yield was as high as 87.5%. Regarding the stimulus response of the sensory organs, it was confirmed that the responsiveness decreased as the frequency increased.
抄録全体を表示
-
世良 一成, 山野辺 夏樹, Ixchel G. Ramirez-Alpizar, Zhenting Wang, 万 偉偉, 原田 研介
セッションID: 1P1-J12
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a robot assembly planning method by automatically reading the graphical instruction manuals design for humans. This method generates an Assembly Task Sequence Graph (ATSG), which is a network structure that defines a series of assembly tasks, including the motions of each object that satisfy the constraints. Based on the proposed method, an ATSG for assembling a chair is generated, and finally the chair assembly task is done by the robot.
抄録全体を表示
-
田 有烈, 延 剛, 船橋 賢, シュミッツ アレクサンダー, 菅野 重樹
セッションID: 1P1-J13
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Most of robotic manipulation tasks are time-related process so it is important to handle time-series information for better manipulation performance. In this paper, we propose a LSTM based prediction system which predicts grasping outcome using time-series 3-axis tactile information. Tactile data was collected from grasping and early post-lifting stage using 25 objects. For evaluation, K-fold cross validation was used considering relatively small size of datasets. Also, ROC curve and PCA were used for analysis. As a result, the proposed model achieved 77.4% prediction accuracy and it performs better than a comparison model which does not utilize time-series information.
抄録全体を表示
-
高橋 晃, 佐藤 将貴, 並木 明夫
セッションID: 1P1-J14
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, research and development of manipulators equipped with a multi-fingered robot hand as an end-effector have been carried out to perform delicate and dexterous tasks. In high-speed movements of such a multifingered hand-arm, the weight of the multi-fingered hand slows down the response of the arm. To solve the problem, we propose a control method to use a high-response hand to compensate for the delay in the low-response arm, for ball throwing. In particular, the accuracy of the pitching motion is improved by predicting tracking errors based on the arm dynamics and using model predictive trajectory generation to compensate for arm tracking errors by the hand motion. Experimental results of ball throwing are shown.
抄録全体を表示
-
小林 聖英, 高橋 瞳和, 内田 敬久
セッションID: 1P1-K01
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this research, we propose an AGV with a simplified mechanism and structure using “Karakuri” technology. The developed AGV can be driven and steered by a DC motor. The crank mechanism, as steering, is equipped with a variable rod length rod that adjusts the timing of turning. An angle of front tire and a trajectory of AGV were evaluated. The estimated values by the simulation and measured values were qualitatively correct. However, difference between estimated and measured values are observed. It is considered that this is largely affected by the friction with the wheel and the gap between the shaft and the shaft hole in the mechanism, which are not included as conditions in the estimated value. According to the results, it was confirmed that proposed the steering with the crank mechanism is useful.
抄録全体を表示
-
澤入 良樹, 李 沛譲, 田島 利基, 朱 赤
セッションID: 1P1-K02
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Currently, the demand for logistics personnel is increasing due to the rapidly growing of online shopping. Workers often have to load and unload packages in a low position. The burden on the workers’lower back is large and consequently it causes heavy back pain of the workers. In this research, we developed a new type of electric powered lift cart that can automatically keep the package height on the cart table to a position that is the most comfortable for workers’loading and unloading. Moreover, we developed a lift mechanism for the cart in which the driving force is constant so that using a comparatively low power motor is possible.
抄録全体を表示
-
坂田 崚真, 武本 純平, 坪内 孝司
セッションID: 1P1-K03
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this paper, we described the development of an object moving robot at a construction site using the mutual position estimation method of multiple omnidirectional mobile robots using only internal sensors. Today’s material transportation requires multiple personnel and is a repetitive and frequent task. To solve this problem, multiple small robots are used for transportation. By properly moving each robot restrained by the material by hand, mutual position estimation between the robots becomes possible, and it is possible to move by operating the controller. In this paper, we conducted an experiment to estimate the mutual position between the robot and the center of rotation using an internal sensor, and verified how the accuracy is affected by the difference in the road surface.
抄録全体を表示
-
山中 雄太, 伊藤 文臣, 佐藤 広都, 中村 太郎
セッションID: 1P1-K04
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this study, the authors started to develop a robot that can clean grease dust in ventilation ducts. In this paper, we investigated the required specifications of the robot, constructed a model of the cleaning brush, and developed grease dust samples. We also conducted cleaning experiments to confirm the cleaning performance of cleaning tools with different bristle and stiffnesses. Form the experiments, we confirmed that the contact area of the brushes affected on the cleaning performance and the cleaning performance of the nylon brush was 68%, which was the highest compared to other brushes.
抄録全体を表示
-
金田 龍希, 井上 浩行, 曽利 仁, 杉本 大志
セッションID: 1P1-K06
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The purpose of this study is to develop a remote controlled small beach cleaning robot using a portable terminal. The developed beach cleaning robot consists of a moving mechanism, a collecting mechanism and a control system. A crawler system was adopted for the moving mechanism to improve traveling stability. In this paper, we carried out the experiments using Bluetooth communication method from the portable terminal. The effectiveness of the developed beach cleaning robot is verified by the experiments.
抄録全体を表示
-
姜 民柱, 原田 宏幸
セッションID: 1P1-K07
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper presents simulation results for grid-based exploration methods using parent and child robots with different functions. Child robot explores sub-map that parent robot can’t explore or the one that need to be explored precisely. Parent robot explores the other map and also carries the child robot to the sub-map. Parent-child system allows application with different functions or types of robots about exploration. Furthermore, we propose a Grass-fire algorithm based Coverage Path Planning algorithm using distance value of the map also used in path planning, and validate the effectiveness of this algorithm.
抄録全体を表示
-
飯田 悠貴, 三木 理, 坂本 航羽, 松日楽 信人
セッションID: 1P1-K09
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Among rampant COVID-19 spread, our way of life has changed. With contactless business operations becoming popular, robots can play a pivotal role in some industries. Especially, it is believed that teleoperated robot has great capacity to adopt to contingent environmental changes. However, visual information from camera on the robot is so meagre that performing pragmatic patrol tasks may become time-consuming even adept operator. In this paper, we propose function which will be of help to patrol a facility efficiently and safely. On basis of empirical manner, requirements for patrol tasks were enumerated and extracted desirable functions. Then, the effectiveness of function was assessed by university’s security officer and improved with their reviews.
抄録全体を表示
-
大竹 將太, 川高 光太郎, 日高 浩一
セッションID: 1P1-K10
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes an obstacle avoidance control for multi robots. The control use an imitative learning and reinforcement learning. The imitative learning is learned by Optimal Reciprocal Collision Avoidance, and the reinforcement learning is learned based on e-greedy. Simulation results via 4robots are shown in order to validity the control.
抄録全体を表示
-
八重本 直希, 前山 祥一, 小柳 栄次
セッションID: 1P1-K12
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, the spread of online shopping has increased the opportunities for home deliveries. The accompanying shortage of personnel and long workinghours have also become social problems. Research and development of transportation robots are expected as a solution to these problems. In our laboratory, a half drone inverted pendulum type transportation robot are also developed. However, the primary prototype in the previous research has less safety and controllability. This paper proposes the second prototype to improve the above weak points and evaluates basic performance of the loading platform with rotors.
抄録全体を表示
-
熊谷 優佑, 山下 貴仁, 妹尾 秀人, 佐藤 恵太, 山本 祐一郎, 田崎 良佑
セッションID: 1P1-K13
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this study, we describe optimization of high-speed vibration control transfer and suppression of residual vibration in a gantry loader. Society requires space saving, low cost, and high speed. Since there has been previous research on space saving, this time we will focus on low cost and high-speed transportation. Currently, the integrated control of AMD and the pre-shaping method can be reproduced by simulation. However, there are some problems in reproducing the simulation in reality. We will solve the problem and show the merits of using both AMD and pre-shaping methods. Ultimately, the goal is to realize integrated control of AMD and the pre-shaping method.
抄録全体を表示
-
多加谷 一輝, 伊藤 文臣, 佐藤 広都, 渡邉 淳一, 中村 太郎
セッションID: 1P1-K14
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper describes the results of an inspection test in an actual environment using a wave propagation type pipe inspection robot, PI-Ro II. it has been difficult to inspect pipes with thin, long, and complicated layouts from the viewpoint of both propulsive force (force to push the robot deeper into the pipe) and traction force (force to pull the robot into the pipe). Therefore, we developed PI-Ro II, a compact robot with a mechanism that enables both propulsion and traction forces, and confirmed its inspection performance in a mockup. However, it is difficult to perfectly reproduce the piping conditions actually used in factories with this mockup. To solve this problem, we conducted an inspection test using PI-Ro II in a pipe used in an actual factory, and succeeded in acquiring images of the inside of the pipe and inspecting 3.8 m of pipe, including passage through two consecutive 90-degree bends.
抄録全体を表示
-
入江 清, 鈴木 太郎, 原 祥尭, 吉田 智章, 友納 正裕, 西村 健志, 大和 秀彰, 清水 正晴
セッションID: 1P1-L01
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper describes our autonomous navigation system based on Laikago, a commercially available quadruped robot. We mounted a 3D Lidar and implemented navigation software suite on the robot to walk according to abstract (non-physical) velocity commands. While many of the equipped software components remain unchanged from those for our wheeled robots, we have newly developed data-driven controller for path following to take advantages of the omnidirectional walking capability of the robot. Our robot system successfully completed 1.5 km of autonomous navigation in our university campus.
抄録全体を表示
-
小木曽 里樹
セッションID: 1P1-L02
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposed a method of estimating three-dimensional location and pose using asynchronous acoustic beacons and microphone array. The method was evaluated with sound propagation simulation in a reverberant room. The reverberation time 60 (RT60) of the room was changed from 0 to 1.5 s. The proposed method was compared with conventional method using deterministic TDOA detection. The results showed the proposed method is more robust against higher RT60 comparing to the conventional method.
抄録全体を表示
-
秋庭 啓吾, 鈴木 龍紀, 池 勇勳, 梅田 和昇
セッションID: 1P1-L03
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this paper, we propose a novel ICP-SLAM approach that can build robust map information even in indoor environments with people. In general, when a mobile robot constructs a map in an environment where there are people nearby, it is difficult to construct a map with high accuracy due to occlusion caused by people. Therefore, YOLO is utilized to detect people in image data and human removal process is applied to improve the accuracy of map information.
抄録全体を表示
-
田所 功, 片岡 良介, 鈴木 龍紀, 池 勇勳, 梅田 和昇
セッションID: 1P1-L04
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this paper, we propose a method to estimate robot pose accurately by considering environmental features for the particle filter. Particle filters have been widely used for pose estimation of mobile robots. However, the accuracy of the self-positioning estimation is degraded in the environment where the shape features are scarce. Therefore, a novel likelihood function is designed by taking into account the environmental features for the particle filter. We improve the accuracy of the pose estimation by fusing the likelihood of the shape information and the likelihood of the binary attribute information in the water puddle for the particle filter.
抄録全体を表示
-
韓 範洙, Ankit A. Ravankar, 江丸 貴紀
セッションID: 1P1-L07
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Recently service robots are increasingly been deployed for different applications such as for homes, hospitals, and other service industries. For efficient navigation of mobile robots in complex indoor and outdoor environments, it is essential for the robot to perform collision-free and optimized planning in dynamic scenarios. Recently, reinforcement learning methods, a type of machine learning technology are used for the navigation tasks in mobile robots and is actively been researched. It has several advantages that the robot can learn and adapt to new and unknown environments quickly by training over several iterations of different scenarios. In this paper, we propose a navigation policy for a mobile robot equipped with a 2D-LiDAR sensor based on the Proximal Policy Optimization (PPO) of a stochastic approach. Our method also includes a stochastic operation which reduces the computational cost of a training session. We also trained our robot in multiple virtual environments to speed up the learning model ’s training. For this research, the navigation test was executed only in a virtual simulator.
抄録全体を表示
-
福田 健太郎, 中村 恭之
セッションID: 1P1-L08
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
We propose AT-COMPL-VAE (VAE with autocompletion feature), which is an extension of VAE (variational autoencoder) to complement 2D LiDAR data and extract features suitable for self-localization. AT-COMPL-VAE is a multi-task neural network with three neural networks: autocompletion, which completes the missing 2D LiDAR data; encoder, which compresses the input data; and decoder, which restores the original input data from the compressed data. In the simulation experiments conducted to verify the effectiveness of the proposed method, we evaluated the performance of 2D LiDAR data complementation and the performance of self-localization using the complemented data. It is demonstrated that the data complemented by the proposed method is close to the true value compared to the data complemented by random values and that the success rate of self-localization using the complemented data was improved by more than 80% compared to the case using the data with missing data due to occlusion.
抄録全体を表示
-
大塩 晃平, 井上 椋太, 辻元 誠, 谷口 和彦, 久保田 直行
セッションID: 1P1-L09
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, labor shortages caused by the declining birthrate and aging population have become a serious problem in various industries. Especially in the construction industry, illuminance measurement work is very hard, so automation of the work is hoped. In past research, we have developed the illuminance measurement robot which moves autonomously and have shown its effectiveness. SLAM used for autonomous movement has big problems such as fail caused by accumulation error and position lost. Our study also confirmed fail of SLAM at several construction sites. In this paper, we propose an evolutionary localization method using CAD map, and discuss its effectiveness from the viewpoint of improving the efficiency of illuminance measurement work through multiple running experiments.
抄録全体を表示
-
湯浅 連, 横山 光希, 森岡 一幸
セッションID: 1P1-L10
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The purpose of study is autonomous mobile robot navigation in any unknown real environments. A proposed system is based on switching of action models according to situations around the mobile robot. Action models for predefined basic shapes are trained in advance by deep reinforcement learning. In addition, mobile robot selects suitable action model based on shapes around the robot. Autonomous robot navigation experiments were performed in two real environments that include the place mainly consisted of basic shapes and the other place including complicated shape. The results of the experiments are shown in this paper.
抄録全体を表示
-
松原 佑樹, 末永 良太, 森岡 一幸
セッションID: 1P1-L11
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The purpose of this study is that multiple mobile robots travel to their destination without interfering with each other. This paper introduces an implementation example of cooperative navigation using a network of two mobile robots. The proposed system uses Rowma, which is a robot network construction system, to share position information among multiple ROS-based robots. According to position relationship among robots, a route to avoid the other robot is selected. In the experiment, the robot was able to reach a destination along a route avoiding the other robot. The result shows that it is possible to achieve autonomous navigation without interference between mobile robots by using Rowma.
抄録全体を表示
-
梅山 隆介, 新島 駿, 佐々木 洋子, 竹村 裕
セッションID: 1P1-L12
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The paper presents a graph design for action planning of autonomous mobile robots. Mobile robots, especially in city situations, are required various actions to move to their destination. When moving around in a building, it is not only about planning the wheels’ path, such as opening doors to get inside or getting on elevators, but also about the actions other than movement. This paper proposes "the action node graph" to realize action planning that includes the actions that occur as they move. The action node graph is dynamically generated from the action information required at each location and the robot’s possible actions. The robot searches their route plan that includes behaviors other than movement. The search results are automatically converted into a Behavior Tree to realize the mobile robot’s autonomous action.
抄録全体を表示
-
井上 一道, 藤倉 駆, 小島 拓登, ウメイニ マイケル オネディカチュク, 清水 美羽, 牧野 空知, 齊藤 星南
セッションID: 1P1-L13
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The method that combines Pure Pursuit Algorithm and Dynamic Window Approach (DWA) is effective for realizing stable running of the robot. However, DWA has the problem that it is caught by the local minima when surrounded by obstacles. In this study, we examine the solution of the problem by effectively selecting the tracking point of Pure Pursuit Algorithm. The experimental results in the real environment showed that the proposed method works effectively.
抄録全体を表示
-
中井 勝也, 鄭 聖熹
セッションID: 1P1-L14
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
It is difficult for mobile robot to estimate accurate localization by odometry. In particular, it is difficult for omni-directinal mobile robot to estimate accurate localization by odometry. Position error of mobile robot is mainly because of slips on the wheels. Therefore, each angular velocity is compensated the error of the velocity of the mobile robot that calculated by introducing the error coefficient. The purpose of the research is to aim for autonomaous movement and map generation in an environment including dynamic obstacles using omni-directional mobile robot with mecanum wheel. In this paper, we investigate in order to give an error confident, we consider the tendncy of the measurement error of the same mobile robot with mecanum wheel in the actual environment.
抄録全体を表示
-
原 祥尭, 岡田 佳都, 赤井 直紀, 横塚 将志, 冨沢 哲雄, 伊達 央, 黒田 洋司, 坪内 孝司
セッションID: 1P1-L15
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper describes the online activities of Tsukuba Challenge 2020. Due to the spread of coronavirus (COVID-19 pandemic), the experiments where all teams gathered in Tsukuba were canceled. As an alternative, we proposed the following activities: navigation experiments at each team’s local site, provide and use of datasets, and use and development of simulators. In addition, to keep the community connections, an orientation meeting, lightning talks, and a symposium were held online.
抄録全体を表示
-
軍司 健太, 大野 和則, 岡田 佳都, 小島 匠太郎, 昆陽 雅司, 田所 諭
セッションID: 1P1-L16
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The authors aim to build an autonomous system that works in factories and shopping malls. Objects layout are designed for supporting human activities in such environments. This paper proposes a method of SLAM using objects layout constrains. The proposed method improves the accuracy of the objects layout map by using the objects layout constraints. The function is verified using the simulation and the effectiveness is confirmed from the numerical evidences.
抄録全体を表示
-
―地物の方位と距離を利用した特徴量による位置推定と道なり走行によるナビゲーション―
品田 直樹, 四宮 雄平, 天野 嘉春
セッションID: 1P1-L17
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This study aimed to apply the essence of human navigation methodology to autonomous vehicles with an abstracted schematic floor map image without using a high-precision pre-map such as SLAM. First, we transformed the captured floor map image into a cognitive feature map represented by cognitive features. Self-position estimation and path planning should be done efficiently with the cognitive feature map. Next, we converted the sensing data of the moving object into cognitive features and mapped the position and posture of the moving object in the moving space to the position and posture in the cognitive feature map. In addition, after extracting the branches to be passed to the destination using the attribute information obtained from the floor map, we appropriately placed the sub-goals in the moving space and drove along the road. In the virtual environment, a virtual vehicle equipped with a stereo camera and an encoder was used for navigation using the proposed method. Even if the self-position estimation is inaccurate temporally, the proposed method does not interfere with the navigation because it controls the vehicle by running along the road.
抄録全体を表示
-
武井 美樹, 大坪 義一, 坂田 誠一郎
セッションID: 1P1-L18
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, due to the increasing size of buildings and the concentration of urban areas, the damage caused by disasters such as earthquakes tends to increase. In this case, deploying rescue robots has the effect of reducing the damage caused by secondary disasters. Therefore, in this study, we develop a method for a robot to autonomously construct a three-dimensional map and determine its search path, assuming the GPS function cannot be used in the environment. Heights at several locations are measured in a target environment, and the 3D map considering the height distribution of obstacles is constructed by the Kriging method. The path search method based on the potential method is applied to the height map, and a moving path avoiding obstacles is determined by the proposed approach.
抄録全体を表示
-
平賀 拓実, 山田 竜平, 矢口 勇一
セッションID: 1P1-L19
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The 3-D mapping data are necessary information for autonomous robot navigation. We progress a study to construct a 3-D point cloud map with extensive range and high resolution using two LIDARs which cover horizontal and vertical ranges. In this paper, we target construction of a calibration method to obtain parameters of coordinate conversion to integrate two LIDAR’s point cloud maps with good accuracy. In our method, we prepared for some types of boards where circles are opened and measured the boards using the two LIDARs simultaneously. The parameters of coordinate conversions are determined to match the circles on boards in two LIDARs data. Then, we have found that usage of multiple boards which multiple circles with different size and oblique array are opened is useful to obtain the good parameters.
抄録全体を表示
-
橋本 雄紀, 石塚 裕己, 池田 聖, 大城 理
セッションID: 1P1-M02
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, soft tactile sensors comprising liquid metal and silicone rubber have been studied for tactile sensing of soft robots. In order to obtain detailed information about objects, the sensor array was configured by utilizing crossing of orthogonal microchannels filled with liquid metal as sensing elements in previous research. In this study, we propose the soft tactile sensor array with independent sensing elements. This sensor embeds narrow microchannels standing vertically in silicone rubber. Microchannels are filled with liquid metal and used as sensing elements. This structure realizes miniaturization and high-density placement of sensing elements while maintaining the independence of the sensing elements. We fabricated the sensor with 16 sensor elements of 0.4mm width embedded at 2.0mm intervals.
抄録全体を表示
-
相良 桃子, 竹村 研治郎
セッションID: 1P1-M03
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Tactile sensation is an important factor for products to get added value. Therefore, tactile sensing attracts huge attention as quantitative evaluation of tactile sensation. While human tactile sensation recognition mechanism has nonlinearity, many of previous studies are confined to construct linear regression models to estimate a tactile sensation. Hence, this study proposes a non-linear tactile estimation model of molded plastic plates. We extracted features from vibration information mediated by mechanoreceptors’ perceptibility, and the relationship between the tactile evaluation score and extracted features was formulated by constructing regression model including non-linearity. The result implies what range of vibration information affects each tactile sensation.
抄録全体を表示
-
大髙 峻, 中山 昇
セッションID: 1P1-M04
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In this study, in order to clarify the electrical conduction mechanism and F-R characteristics of a new conductive material, Silicone/VGCF, which can be used as a force sensor and a garment pressure sensor, we clarified the effect of the volume content of VGCF on the electrical properties. As a result, it was found that the electrical properties of silicone/VGCF changed due to the change of the conductive path inside the molding depending on the VGCF content and the load. Furthermore, it was suggested that Silicone/VGCF could be used as a force sensor and a garment pressure sensor.
抄録全体を表示
-
牛山 奎悟, 髙橋 哲史, 梶本 裕之
セッションID: 1P1-M05
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
The purpose of this study was to investigate whether stimulation of the proprioceptors in the arm during active movement can affect not only the proprioception of the arm but also the perception of the hand-held object. If it is possible to control the perception of a hand-held object through stimulation to the body, it can be applied to virtual-reality interfaces and controllers, which can be used in a wide range of situations. In the experiment, participants were given a context in which they grasped an object and actively moved. Proprioceptive stimulation was applied to participants using vibration and skin stretch based on the kinesthetic illusion. They were asked to evaluate the perception of the object and the arm. Consequently, it was found that the perception of not only the arm but also the hand-held object could be changed, although there were large individual differences.
抄録全体を表示
-
臼井 達也, 石塚 裕己, 池田 聖, 大城 理
セッションID: 1P1-M06
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Tactile sensors are necessary for robots to understand surrounding environments and their states. Recently, soft tactile sensors have been widely developed to integrate them into soft robots. Although, the detection parts and substrates of many soft tactile sensors compose of soft materials such as silicone rubber and liquid metal, stiff components such as electrical elements and wires are also embedded. The stiff embedded components reduce flexibility and durability. In this research, we develop a soft capacitive tactile sensor using the movement of interface between encapsulated air and water in a silicone rubber body. A force applied to the tactile sensor causes a change in the position of the interface. Displacement of the interface changes the capacitance of the sensor. The structure does not require an ohmic connection between a contact and a detection part and improves durability.
抄録全体を表示
-
川島 みなみ, 北村 知也, 浜名 竜大, 境野 翔
セッションID: 1P1-M07
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Functional electrical stimulation (FES) is a method of moving joints by applying electrical stimulation to muscles. Recently, FES has been attracting attention not only as a rehabilitation method but also as an application for healthy people. People use their fingers to perform various tasks, thus enabling them to control their fingers using FES is a critical issue. However, because the muscles that move the fingers are located complexly in the forearm, joint angle control has not been conventionally performed with five fingers. In this paper, we focused on the fingers, which are especially necessary for detailed work, and controlled their joint angles. The stimulation positions in the forearm were identified using multi-point electrodes. The control was performed by modeling the relationship between voltage and joint angle. Although the proposed method drove not all the subjects’ five fingers, the fingers whose stimulus positions could be identified could follow the target value.
抄録全体を表示
-
菊池 大輝, 昆陽 雅司, 田所 諭
セッションID: 1P1-M08
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Phantom sensation (PS) can be used as a method of giving vibrotactile stimuli to various positions on the body with fewer actuators. The conventional method assumes only a simple sinusoidal waveform to calculate the parameters for PS generation, but it is not applicable to arbitrary waveforms such as audio signals and collision vibrations. In this study, we propose a method to generate PS based on the vibration intensity perceived by humans. The vibration intensity is calculated using Intensity Segment Modulation (ISM). The experiment showed that the proposed method could properly generate PS and it can be used regardless of the distance of two vibrators less than 8 cm.
抄録全体を表示
-
亀岡 嵩幸, 梶本 裕之
セッションID: 1P1-M09
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
In recent years, many researchers have been investigating the addition of other modalities such as haptics to the visual and auditory based VR experience using HMDs. On the other hand, these haptic devices are cumbersome to wear and their fingers cannot be moved freely. We solved these problems by embedding a suction tactile presentation device in an HMD and presenting the finger sensation in a VR environment to the face. In this paper, we investigate how changes in the suction hole and suction pressure affect the tactile sensation. As a result, we found that a dense arrangement of 2 mm suction holes can provide strong tactile sensation even with weak suction pressure, and that pressure sensitivity is also easily presented.
抄録全体を表示
-
浦宗 龍生, 石塚 裕己, 平木 剛史, 川原 圭博, 池田 聖, 大城 理
セッションID: 1P1-M10
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
Pneumatic actuators have been applied to wearable haptic devices owing to their lightweight. However, the pneumatic actuators require an air compressor and air valves. Thus, the resulting systems are large and heavy. We have studied a haptic display using a liquid-to-gas phase change actuator. The actuator is composed of a thin plastic pouch containing a low boiling point liquid. A Peltier device controls the temperature of the liquid in close contact with an actuator. In this study, we embedded a small pressure and temperature sensor in the actuator and measured the pressure change induced by heating. The result indicates that the maximum pressure in the actuator is determined by vapor pressure, and a few tens of seconds are required to reach the maximum pressure.
抄録全体を表示
-
~第1報 キャリブレーションに関する基礎理論の構築~
段上 将門, 河村 晃宏, 倉爪 亮
セッションID: 1P1-M11
発行日: 2021年
公開日: 2021/12/25
会議録・要旨集
認証あり
This paper proposes a new bolt type force sensor that measures forces applied to an object fixed by the sensor. This sensor consists of a general metal bolt and four strain gauges. Due to the feature of the bolt type sensor, it can measure the force applied to the object without extra jigs and gaps. In this report, the structure of the sensor is introduced and its calibration strategy is discussed. Finally, a force measurement experiment is conducted.
抄録全体を表示