著者

青木 真也

出版者

法政大学大学院理工学・工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.58, 2017-03-31

---

Polystyrene block copolymers containing triethoxysilyl (TES) groups were synthesized by a living anionic polymerization and the subsequent functional group transformation reactions. Then thin polymer films were applied for the reaction with organic acid to form Si-O bonds at the film surface. Surfaceconcentration of Si was confirmed by X-ray photoelectron spectroscopy (XPS) measurement for the resulting polymer film. In addition, contact angle of water droplet for the film increased after acid treatment.These results clearly showed that the polysiloxane substrate was prepared at the outer most surface by treatment with organic acid.Key Words : living anionic polymerization, triethoxysilyl group, surface characterization, block copolymer

著者

成田 結香

出版者

法政大学大学院理工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.60, pp.1-7, 2019-03-31

---

Laparoscopic surgery is a difficult surgical procedure as compared with open operation. Therefore, training is necessary for surgeons before conducting laparoscopic surgery clinically. In this paper, improvement and extension of the training system which has been developed by our group was carried out. First, effectiveness of the evaluation method for trainee's hands movement was verified through the past measurement data obtained for operation of ligation. Second, in order to make it easy to perform ligation under Para-axial position, camera position of the training box was improved. Third, former sensor module which was attached to the forceps was downsized so as not to collide to other objects. Finally, the proposed evaluation system for hands movement and the evaluation system for angle of the wrists which was developed in our previous study were merged. Then, the evaluation results of trainee's angle of the left and the right wrists, hands movement and operation time in comparison with those of the excellent skilled surgeon are displayed on the monitor screen.

著者

市川 竜太郎

出版者

法政大学大学院理工学・工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.62, pp.1-6, 2021-03-24

---

Have you ever imagined the character and gender of the writer by looking at the handwriting ? There has been a lot of research on the relationship between handwriting and personality for a long time. However, there may be some changes in Japanese handwriting with the times. If that is the case, we may find something new in Japanese handwriting nowadays. The purpose of this study is to explore the structure of the impressions of Japanese handwriting using SCT.

著者

四木 宏香

出版者

法政大学大学院理工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.61, pp.1-6, 2020-03-24

---

The multi-map navigation using SLAM is one of the current research topics of the mobile robots. In this paper, we describe a study of a mobile robot navigation using the multi-map method. By applying SLAM, both the map and trajectory of a mobile robot can be generated simultaneously. However, the drawback of SLAM depends on sensor accuracy, which may cause mapping failure. One of the solutions to prevent the failure of mapping is not monolithic mapping but short-term multiple mapping. The advantage of short-term multiple map generation can reduce map construction failures. The other typical applications for multi-map is multi-floor navigation. In this paper, we apply multi-map method for both navigations. The usefulness of the proposed method was confirmed by simulation and actual experiments.

著者

河野 辰哉

出版者

法政大学大学院理工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.60, pp.1-6, 2019-03-31

---

This paper describes the development of Robot Operating System (ROS) component for lane recognition and navigation of IGVC Auto - Nav Challenge. To achieve a robust and stable navigation, we propose new lane and obstacle recognition algorithm based on omnidirectional camera. Employing fast NCC based piecewise template matching technique enables robust lane detection regardless of surrounding brightness changes and various lane shape including sharp curves. To recognize surrounding obstacles only vision sensor without LiDAR, we employ YOLOv3-tiny based deep learning algorithm which can avoid obstacle collision. Since ROS has capability of environmental simulation by using Gazebo, rapid prototyping is achieved by applying both simulations and actual experiments which can significantly reduce development period. The effectiveness of the newly proposed algorithm was confirmed by both offline simulation and actual outdoor experiment.

著者

伊藤 千鶴

出版者

法政大学大学院理工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.59, pp.1-4, 2018-03-31

---

We can imagine that one having more useful information gets superior than others if they are in the same condition and under some struggling situation. For example, in a supermarket, we naturally choose the shortest queue of cashiers in order to finish our payment with minimal waiting time. If one knows a better strategy to choose the queue, she or he can finish the payment in more shorter time. However, the more customers know the strategy, the less the effectiveness of using the strategy. In this study, we assume one strategy in such a situation, and investigate its effectiveness via simulation study.

著者

蔡 詩祐

出版者

法政大学大学院理工学・工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

vol.56, 2015-03-24

---

Goto and Shibata(2010) proposed a reinforcement learning algorithm using a recurrent neural network. Back Propagation Through Time (BPTT) was used for the neural network’s learning rule. This algorithm autonomously acquires prediction functions for tasks that are difficult to be accomplished without these functions. To verify the effectiveness of this method, they used episodic tasks where the starting state and the terminal state of the tasks could be given explicitly. However, in the real world, there are many continuous tasks that cannot indicate the starting and terminal state in detail. This study verifies the performance of the previous method on continuous tasks and presents a new method that uses Real Time Recurrent Learning (RTRL) which allows real-time learning for the neural network. The results indicated that the previous method had a good performance even in continuous tasks. On the other hand, the new method using RTRL was inferior to the previous method in performance for both continuous tasks and episodic tasks.

著者

横林 亮平

出版者

法政大学大学院理工学研究科

雑誌

法政大学大学院紀要. 理工学・工学研究科編 (ISSN:21879923)

巻号頁・発行日

no.60, pp.1-8, 2019-03-31

---

The tensor data model is capable of detailed data representation and is compatible with data mining. However, there is a problem that the data volume becomes enormous and processing time is required. It is one solution to this problem to improve the efficiency of a series of data processing such as insertion, deletion, search of data in the secondary storage area. In this paper, we propose an serial search method using locality sensitive hashing. The hash technique is a search method that is hardly affected by the data amount, but there is a problem that the serial search can not be performed. By using locality sensitive hashing, we sequentially arrange similar data in the same block and improve efficiency by carrying out parallel processing for each block. We will also describe the efficiency of tensor data manipulation (especially data mining manipulation). Using locality sensitive hashing, realize high-speed multidimensional association rules extraction limited to partial data.