2022 Fiscal Year Final Research Report
State understanding of onboard cameras via high-speed proximity road surface analysis
| Project/Area Number |
20K19891
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 61050:Intelligent robotics-related
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | 車載画像処理 / 高速ビジョン / 高速路面解析 / 自動キャリブレーション / 環境認識 |
|---|
| Outline of Final Research Achievements |
The objective of this study was to obtain a fast, accurate, and robust estimation of the state of a moving vehicle's onboard vision system, such as its attitude and the amount of movement, using high-speed proximity road surface analysis, which analyzes changes in the appearance of textures on the proximity road surface using a high-speed camera. Results showed that the system was able to estimate the attitude angle within 1 degree and the velocity within 1% even in situations where the camera was vibrating significantly. Furthermore, by applying the high-speed proximity road surface analysis, an estimation method for the attitude angle of a typical vehicle-mounted camera relative to the vehicle coordinate system was proposed, and its estimation error was shown to be within 1 degree. In addition, highly accurate distance measurement using stereo high-speed vision and a high-speed detection method for pedestrians jumping out of blind spots were developed as application examples.
|
| Free Research Field |
コンピュータビジョン
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究により移動体に搭載したビジョンの動的な状態の把握が可能となるため，移動体自身や周囲物体の位置・速度を計測する精度や速度を向上させるための基盤的な技術を提供する．これにより，自車と周辺環境との位置関係の高速高精度なセンシングが要求される応用において特に有効である．本研究で直接の対象とした自動車だけでなく，AGVや列車，ドローンといった移動ロボット全体に対しても応用可能な技術であるため産業的な価値の高い技術であるといえる．
|
