2023 Fiscal Year Annual Research Report
A Platform for Hierarchical Data-Driven Design, Fabrication, and Control of Modular Soft Robots with Slender Beams for Locomotion and Manipulation
| Project/Area Number |
23H01376
|
|---|
| Allocation Type | Single-year Grants |
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
タ デゥックトゥン 東京大学, 大学院情報理工学系研究科, 助教 (20869226)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
千葉 滋 東京大学, 大学院情報理工学系研究科, 教授 (80282713)
川原 圭博 東京大学, 大学院工学系研究科(工学部), 教授 (80401248)
|
|---|
| Project Period (FY) |
2023-04-01 – 2027-03-31
|
| Keywords | Body Design / Modularization / Morphology / Soft Sensors / Soft Robots |
|---|
| Outline of Annual Research Achievements |
In this fiscal year, our research focused on two sub-projects:
Sub-project 1: Single-Segment Flexible Beam for Undulation Locomotion: We designed a tapered, pre-compressed flexible beam to achieve undulation motion using a single actuator. This project provided valuable insights into the behavior of variable stiffness flexible beams under different pre-compressions when driven by a motor. The result of this project was submitted to IROS 2024.
Sub-project 2: Automation of Body Design for Two-Segment Snake-Like Soft Robots: We implemented a genetic algorithm to explore the design space of anisotropic frictional skins for snake-like soft robots. This innovative approach enables us to optimize the body design of these robots for improved performance and functionality. A greedy version of the algorithm was presented at the 86th IPSJ National Convention. Based on this, we summarize the design pipeline into a poster that is presented at the ICRA 2024 RSA Workshop. We also implemented a more sophisticated version of the anisotropic frictional skin optimizer that considers both the magnitude and direction of the frictional vector on the designed robotic skin. The optimizer is parallelized by using multiple computational cores on a HPC cluster.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Enabled by newly invested equipment such as an SLA printer, a motion capture system, and a high-performance computing infrastructure, we could experiment seamlessly with faster iteration. This helps to accelerate the data collection and shorten the prototyping cycle of soft robots.
|
| Strategy for Future Research Activity |
In this fiscal year, besides continuing projects from last year, we will start working on:
(1) Modularization of snake-like soft robots: modularize snake-like soft robots into separately independent flexible beams that can be connected into one longer snake-like soft robot.
(2) Serial flexible beam as a manipulator: instead of locomotion, use a serially connected flexible beam as a manipulator to study multi-segment soft continuum arms.
(3) Reinforcement learning for co-design of soft-bodied robots: improving from the genetic algorithm, we will implement reinforcement learning in body shape optimization for specific tasks.
These 3 projects will continue laying foundation for a platform that help co-designing soft-bodied robots.
|
