| 研究実績の概要 |
In this year, system generalization was developed to work with complex bones shapes and adapt it to clinical conditions. The methodology was completed to simplify the pattern (including the condition that its 2D projections are "star-based polygons"), it was presented at national and international conferences [MIMS2018,JSIAM2018,CMD2018,7OSME]. After discussion with physicians about its adaptation to the clinic, a more accurate representation of the surface of the bones was needed (for example, the epiphysis of the proximal and distal femur). A new pattern design methodology based on the opening of the tree structure of traditional origami patterns was developed to address above problem. Thus, we developed a two arm robot to imitate the human's folding steps and continued developing automatic learning algorithms to recognize hands and paper states in scenes of origami performing (e.g.bending,gluing,sharpening). Videos are used to discover folding policies by robotic reinforcement learning in a simulation environment. Preliminary results with tracking were presented at national conferences [MIMS2019] and will be presented at international conferences [ASME2019, OKINAWA2019]. In addition, we worked with videos of driver's facial expressions in autonomous cars' simulation environments to explore possible forms of coordination between man and machines. To adapt to the clinical conditions, a software prototype was developed to obtain the 3D models from X-ray images in the clinic and the doctors were trained to use them.
|
| 今後の研究の推進方策 |
Currently we are working on the integration of two robots: one that is responsible for building the simple patterns obtained from the decomposition of the 3D parts using "star-based polygons" and another that is responsible for completing or inspecting the more complex parts using two robotic arms. It is necessary to explore possible forms of coordination between the two robots for the development of a common task in critical collaboration environments. There are still problems with the gluing of paper. It is necessary to reduce the number of gluing areas and, by using ANSYS-based simulation, to try other methods to reduce errors. During the discussion with orthopedic specialists, the idea of using the developed patterns to develop a new type of orthopedic nail arose. We are currently in the process of simulation with the new nail design.
|
