Improving Clustering Accuracy for Object Tracking based on DBSCAN and IoU Techniques
自動運転のためのクラスタリングの改善による移動物体追跡のロバスト化
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.):¥1,100 Members (tax incl.):¥880
- Publication code
- 20225008
- Paper/Info type
- Proceedings (Spring)
No.2-22
- Pages
- 1-6(Total 6 p)
- Date of publication
- May 2022
- Publisher
- JSAE
- Language
- Japanese
- Event
- 2022 JSAE Annual Congress (Spring)
Detailed Information
Author(J) | 1) 仲田 尚司, 2) 米陀 佳祐, 3) 菅沼 直樹 |
---|---|
Author(E) | 1) Hisashi Nakada, 2) Keisuke Yoneda, 3) Naoki Suganuma |
Affiliation(J) | 1) 金沢大学, 2) 金沢大学, 3) 金沢大学 |
Affiliation(E) | 1) Kanazawa University, 2) Kanazawa University, 3) Kanazawa University |
Abstract(J) | 移動物体の追跡の精度は,自動運転の安全性に大きく関わる.物体の点群の誤クラスタリングによる誤認識の改善は,追跡の安定化につながる.本研究では,誤って統合された物体を,DBSCANを用いて個々に分割することや,誤分割された物体を,IoUを用いた統合により修正することによる,追跡性能の向上を目的とする. Translation |
Abstract(E) | Accurate tracking of moving objects is very important to ensure safe maneuver of autonomous vehicles. However, this demand becomes very challenging in the dynamic environments where two objects might be combined because of the short in-between distance or an object is miss-segmented into two clusters because of blocking LIDAR laser beams by closer obstacles. In this paper, we investigate these two cases and propose a robust clustering tactic to recover the objects shapes based on DBSCAN and IoU techniques. |