| Project/Area Number |
22K17974
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 61050:Intelligent robotics-related
|
|---|
| Research Institution | Kanazawa University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2022-04-01 – 2024-03-31
|
| Project Status |
Completed (Fiscal Year 2023)
|
| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2023: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Fiscal Year 2022: ¥2,730,000 (Direct Cost: ¥2,100,000、Indirect Cost: ¥630,000)
|
|---|
| Keywords | Graph SLAM / Autonomous Vehicles / Mapping Systems / LIDAR / SLAM in Multilevel Roads / LIDAR Mapping Systems / Largescal Map Generation / 2.5D Maps / 2.5D Elevation Maps |
|---|
| Outline of Research at the Start |
Generating precise maps in multilevel environments is very challenging in cities packed with longitudinal bridges (Tokyo) because of severely obstructing the satellite signals by road structures even though using GNSS/INS-RTK systems. Therefore, the previously proposed Graph SLAM (GS-XY) in 20K19893-00 to generate accurate maps in a single XY-Layer using LIDAR road surface images will be modified to generate precise 2.5D maps using elevation images of the road surfaces with recovering the consistency between layers in the Absolute Coordinate System and enabling to combine and update maps.
|
| Outline of Final Research Achievements |
Generating precise 2.5D maps in challenging environments was mainly addressed in the first year whereas dealing with multilevel road structures to precisely aligning the layered road context in the global coordinate system was the goal in the second year. Thus, the elevation images have been integrated into the mapping system to indicate the altitudinal values the road surface images. The loop closure module has been modified to detect and distinguish the road layers based on the elevation information automatically. Accordingly, the cost-function was developed to optimize the positions of the road surface images in the XY plane and then minimize the elevation errors at the detected loop-closures and ensure the global map consistency and coherency in XY and Z planes. The cost function has then been modified to combine maps in terms of updating the road surface representation, expanding the encoded areas and adjusting the map global position for precise localization in the real world.
|
| Academic Significance and Societal Importance of the Research Achievements |
The proposed mapping system outperformed an accurate GNSS-RTK systems to generate precise 2.5D maps in challenging multilevel environments such as Bejoji and Ohashi junctions and accurately combine maps that collected by different agents to increase the safety and accuracy of autonomous driving
|
