2016 Fiscal Year Research-status Report
整形外科手術前計画に役立つ紙ベースのラピッドプロトタイピングシステムの開発
| Project/Area Number |
16K01422
|
|---|
| Research Institution | Meiji University |
|---|
Principal Investigator |
ディアゴ ルイス・アリエル 明治大学, 研究・知財戦略機構, 研究推進員 (20467020)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
篠田 淳一 明治大学, 研究・知財戦略機構, 研究推進員 (60266880)
|
|---|
| Project Period (FY) |
2016-04-01 – 2019-03-31
|
| Keywords | origami engineering / medical robotics / neural networks / orthopedic surgery / surgery planning |
|---|
| Outline of Annual Research Achievements |
As initially planned, we worked on the creation of 3D models from two X-ray images. We tested and implemented the method in [http://orthocad.iitb.ac.in/xrayto3d/] using images from the Cuban hospital CIMEQ. The printed 3D models were compared with paper models obtained by the algorithm in [1]. The preliminary results were presented in two conferences [7, 15]. We developed algorithms for extracting information from videos [6, 13] for use in machine learning and simulation [12]. As it is not clear in the videos, which are the most important parameters to teach the robot how to fold the paper, we used convolutional neural networks (CNN) for this task. CNNs were compared with our proposed HNNs and the preliminary results were presented at several national conferences [10, 11]. We proposed a new method to generate digital 3D models and crease patterns to create 3D paper models automatically by the robot without learning since the number of steps and manipulations that must be learned to teach the robot is very high. The new method creates a pattern from a 3D object that can be represented by a surface in revolution. We developed two machines that allow folding and gluing the paper automatically [5]. The machines include a new control algorithm based on HNN that reduces the error of the machines and speed-up the creation of the paper models [8, 14, 16]. The machines are being adapted to create complex shapes like the honeycomb [9]. More complex 3D shapes are created from the combination of HNN and IGA during quantification of personal perceptions [2, 4]
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
The results are better than initially planned. We participated in the ASME Student Mechanism & Robotics Design Competition Award 2016, developed in ASME (IDETC/CIE 2016) 21-24 August, Charlotte, USA and achieved the Third Place among more than 100 proposals from all over the world with the work entitled “Norigami Folding Machines For Complex 3D Shapes” by J. Romero, L. A. Diago, J. Shinoda, Chie Nara and I. Hagiwara. The volume of publications and national and international presentations is also higher than initially anticipated.
|
| Strategy for Future Research Activity |
Research 1.1) It was achieved the creation of 3D models from X-ray images following the following steps: 1) Extraction of the contours of the bones 2) Creation of 3D model from implicit functions. However, in order for the method to be used in the clinic, according to CIMEQ specialists, the following problems must be solved: Problem 1: quit step 1 to draw the contour, because it depends on the wrist pressure and that generates errors. Problem 2: current program is not faithful because it obviates some details that are important for guidance at the time of surgery for example the epiphysis in the tibia rod.Research 1.2*) ,1.3*) We will improve the quality of 3D models by using the bone database (http://lifesciencedb.jp/bp3d/) to create X-ray images with ray-tracing of implicit functions and matching of ray-tracing images with real images.Research 3.2) Continue exploring the collaboration between the robot and the human to create 3D models in paper from the following steps: 1) Record videos of people doing origami and play on PC using simulation and 2) Extract the steps of the video and give orders to the robot = Direct teaching.Research 3.3 *) As the information contained in the videos is so large and it is not clear which are the most important parameters to teach the robot how to fold the paper, we must extend the proposed FQHNN method as a deep learning method for this task.Research 3.4) Test the proposed new method with the bones of the database (lifesciencedb.jp/bp3d/) and evaluate the results from the comparison with the existing methods based on polygonal meshes.
|
| Causes of Carryover |
KUKA youBot 2 arms system was initially planned for the development of the robot hands but we developed a prototype of a robot including two machines (called Norigami Machines) designed with LEGO NXT Technology. The current LEGO machines use a width of A6 paper size that limit the size of the constructed 3D paper models. So, a new Norigami Machine able to deal with A4 paper size is under development and the fund will be used in next fiscal year for its construction.
|
| Expenditure Plan for Carryover Budget |
The fund will be used for the construction of a new Norigami Machine able to deal with A4 paper. The created prototype using A4 paper will be introduced in the Hospital for testing and comparison of the models created with previous approach (see paper [1]).
|
-
-
-
[Journal Article] “Norigami Folding Machines For Complex 3D Shapes”2016
Author(s)
[3[J. Romero, L. A. Diago, J. Shinoda, Chie Nara and I. Hagiwara
-
Journal Title
International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Volume 5B: 40th Mechanisms and Robotics Conference
Volume: 5B
Pages: -
DOI
Peer Reviewed / Open Access / Int'l Joint Research / Acknowledgement Compliant
-
-
-
-
[Presentation] “Emotion simulation using emoticons”2016
Author(s)
[12]L. A. Diago, H. Abe and I. Hagiwara,
Organizer
35th JSST Annual Conference International Conference on Simulation Technology (JSST2016)
Place of Presentation
Kyoto University
Year and Date
2016-10-27 – 2016-10-29
-
-
-
-
-
-
-
-
-
