| 研究課題/領域番号 |
20K14713
|
|---|
| 研究機関 | 長岡技術科学大学 |
|---|
研究代表者 |
TRAN PHUONGTHAO 長岡技術科学大学, 工学研究科, 助教 (20848923)
|
|---|
| 研究期間 (年度) |
2020-04-01 – 2023-03-31
|
| キーワード | Motion control / Force control / Intelligent sensing / Deep learning / Robot |
|---|
| 研究実績の概要 |
To build the leading safe, smart and reliable human-robot collaboration system, the superior wideband force observer is proposed to achieve the fine force perception which ensures the precise force perception, especially interaction force with high frequency range. The superior wideband force perception of 1 kHz has been realized. The bandwidth of 1 kHz is much wider than human tactile sensation (400 Hz), and is a record in the field of robot force control. Moreover, the intelligent force sensing has been proposed to realize the high robustness to the variable noise characteristics. The intelligent force sensing has achieved the adaptive force sensation through force observer’s automated parameter tuning under condition of fluctuating noise. The intelligent force sensing has also been designed with the construction of periodicity elimination to solve the problem of undesired and variable periodicities which degrade the performance of force estimation. In addition, the methodology of intelligent force controller for safe human-robot interaction has been being developed together with the intelligent force sensation.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Up to now, the proposed force sensation method has realized the widened force sensing bandwidth of 1 kHz through experimental implementation. The bandwidth of 1 kHz is much wider than human tactile sensation (400 Hz), ensuring the proper force information for controlling interaction with human and environments. Moreover, through experimental implementation, the proposed deep-learning-based intelligent force sensing has realized the adaptive force sensation through force observer’s automated parameter tuning under condition of variable noise. The intelligent force sensation achieves fine interaction force perception under the condition of variable noise characteristics, even in highly noise-contaminated conditions. The periodicity elimination to solve the problem of undesired periodicities which degrade the performance of force estimation has been realized through experiments. The achieved results and the successful implementation of AI based approach promote the completion of the whole intelligent force control system for safe human-robot interaction in the next period of this research.
|
| 今後の研究の推進方策 |
To accomplish the target of this research, the wideband and intelligent force sensation realized will be integrated with the deep-learning-based force controller. The development of the methodology of intelligent force controller for safe human-robot interaction will be continued and realized by experiments in the next year.
|
