知識庫

Dynamic Road Surface Detection Method based on 3D Lidar (3D 雷射雷達動態路面感測技術)
  • 發布年度:2017
  • 主要類別:電動車與車輛電子
  • 次要類別:論文
  • 此論文提出一基於三維光達之道路表面區分之方法,且此方法非必須使用高解析度之三維光達。光達掃描資料在使用靜態校正機制轉換座標後,使用一地面高度估算機制萃取近地點,再內插填補偵測線上之縫隙,而後使用極座標形式分別分析每個點位置之連續性與平滑性找出道路邊緣,最後以區域成長法將建構出道路面。 此方法經兩種不同安裝配置之測試車,進行實車實際道路情境測試,證明此方法之穩定性與效率。

    Abstract—This paper propose a road surface classification method based on 3D-lidar that is not required to be high-resolution. Lidar range data is transformed with static calibration result, and the near-ground points are extracted with a ground elevation estimation method. After filling the scanning gaps with interpolation, the continuity of each scan lines at each point is examined by its polar interpretation. The smoothness is analyzed in the same interpretation with simple linear regression to evaluate the estimated angle feature. The road boundaries defined by the two criteria form the road surface with region growing method. The proposed method has been verified through actual road scenario to show its robustness and efficiency on two different vehicle mounting configurations.

    Keywords—lidar; road; ground; growing; classification