2000 Fiscal Year Final Research Report Summary
Urgent Signal Extraction for Human-interface
| Project/Area Number |
10650262
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent mechanics/Mechanical systems
|
|---|
| Research Institution | Kanagawa Institute of Technology |
|---|
Principal Investigator |
NISHIHARA Kazue Kanagawa Institute of Technology, Department of Engineering, Professor, 工学部, 教授 (60257409)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KAWARAZAKI Noriyuki Kanagawa Institute of Technology, Department of Engineering, Lecturer, 工学部, 講師 (20177752)
|
|---|
| Project Period (FY) |
1998 – 2000
|
| Keywords | voice recognition / urgent signal / emergency / tapping and clapping / sound direction / hand waving |
|---|
| Research Abstract |
In order to issue an order to the machine, we can use some means of crying, voice, tapping or clapping sounds and hand shaking. This report is concerned with recognition of urgent signals at the human-machine interface.
(1) Command by sound or voice : Experiments on catching direction of hand clapping sounds and speaking voice were done by using a speech recognition software(Via Voice (IBM)). The obtained signals were inputted into a 2-degree of freedom servo mechanism which followed the signal directions within accuracy ±15 deg. Next, using wavelet analysis tapping and clapping numbers were recognized surely in a real time under noisy environment. It is confirmed that tapping and clapping sounds are good for urgent signals.
(2) Optical fiber jyro sensor : Combining 3 jyro sensor, we could get very accurate angular velocity signals, which could be a good ordering device to the machine.
(3) Image processing of hand wavings Human can transmit his intention by hand wavings as no-no, bye-bye, nodding and driving away. It is very difficult to distinguish nodding from driving away, however, we should avoid the misunderstanding. To separate motions of nodding from driving away, we introduced a modified DP matching method, and could raise recognition ratio up to 90〜80%. Taking further strict condition, we suppressed missrecognition ratio under 4% though the correct recognition ratio came down to 60%. On the other hand, we distinguished no-no and bye-bye among hand wavings as 100%.
(4) We recommend to use voice control under ordinary conditions, but at the emergency tapping or clapping sounds to stop machines.
|
