| Project/Area Number |
22KF0141
|
|---|
| Project/Area Number (Other) |
22F22769 (2022)
|
|---|
| Research Category |
Grant-in-Aid for JSPS Fellows
|
| Allocation Type | Multi-year Fund (2023)
Single-year Grants (2022) |
|---|
| Section | 外国 |
|---|
| Review Section |
Basic Section 20020:Robotics and intelligent system-related
|
|---|
| Research Institution | Tokyo Institute of Technology |
|---|
Principal Investigator |
中臺 一博 東京工業大学, 工学院, 教授 (70436715)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YEN BENJAMIN 東京工業大学, 工学院, 外国人特別研究員
|
|---|
| Project Period (FY) |
2023-03-08 – 2025-03-31
|
| Project Status |
Granted (Fiscal Year 2023)
|
| Budget Amount *help |
¥2,200,000 (Direct Cost: ¥2,200,000)
Fiscal Year 2024: ¥600,000 (Direct Cost: ¥600,000)
Fiscal Year 2023: ¥1,100,000 (Direct Cost: ¥1,100,000)
Fiscal Year 2022: ¥500,000 (Direct Cost: ¥500,000)
|
|---|
| Keywords | ドローン聴覚 / 音響信号処理 / 深層学習 |
|---|
| Outline of Research at the Start |
This research aims to develop a drone intended for search and rescue. Equipped with "ears", the drone "listens" to an audible target and, in turn, promotes effective communication with them. "Listening" includes recognising, locating and recording the target sound "clearly" (i.e. free of noise from the drone's rotors and surroundings) under harsh, noisy environments, typical in search and rescue.
|
| Outline of Annual Research Achievements |
This research year, we developed and implemented a real-life sound source tracking system using drones equipped with microphone arrays. Previously, such systems were only demonstrated in simulations. Due to strict drone flight regulations in Japan, we designed an indoor system with miniature drones and custom microphones to navigate these restrictions. We also enhanced the system with a drone navigation setup that continuously adjusts drone positions to maximize sound tracking accuracy. Further modifications to the sound tracking algorithms were necessary to address real-life constraints and challenges.
Additionally, we have advanced drone noise reduction techniques for improved sound source tracking, showing promising results in simulations. Real-life testing, however, unveiled unforeseen challenges not anticipated in the simulation phase, necessitating further simulations and experimental tests to refine our approach and meet these new requirements.
These developments contribute to our smart drone audition research theme, utilizing multiple drones with autonomous navigation to enhance sound source tracking performance. This system not only aims to improve the accuracy of locating sound sources but also enhances the quality of recorded audio by minimizing drone noise interference. These advancements are expected to significantly benefit applications where audio clarity and quality are critical.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Given that it is our understanding that until now, no one has developed a multi-drone system that is used to perform sound source tracking, it is under the expectation that developing such a system will be challenging and will take a considerable amount of time. Several components in the system required custom hardware, and many trial-and-errors were needed to overcome the practical challenges and issues that arose, of which many have very limited information or solutions available.
Due to the time spent on developing this system, there has been limited development in the algorithm aspect of the research, which has impacted the research output in terms of publications.
However, most of the challenges have finally been overcome, and we now have a working real-life system to perform the designated task. We now have a development platform to prototype and test any new algorithmic developments of sound source tracking, and even other drone audition-related algorithms. We expect the number of research outputs to increase from this point onwards.
|
| Strategy for Future Research Activity |
We intend to continue the smart drone audition research in the following ways:
1) We intend to make use of the newly developed indoor system to prototype new sound source tracking algorithms, and more importantly drone noise reduction techniques to improve the performance of multi-drone sound source tracking. Such algorithms include namely i) drone noise reduction techniques, ii) sound source recognition (for multiple sound source scenarios, where identifying and ensuring the correct sound source is tracked is important) and if possible, iii) implemented obstacle avoidance and environmental mapping functionalities to improve the drone’s self-reliant capabilities.
2) Using the experience and software developed from the indoor system, we intend to expand and develop a full-sized outdoor multi-drone system for practical sound source tracking. This not only includes development of relevant algorithms, but also design choices in the hardware of the full-sized system. We also intent to integrate the multi-drone sound source tracking system with other forms of robots to allow cooperative sound source tracking to be carried out.
|
