研究目的
Providing semantic information in the form of stationary and dynamic classification by applying a transferable belief model to LIDAR point clouds.
研究成果
The presented approach for semantic motion classification using an evidential combination rule of belief masses is relatively lightweight and provides state-of-the-art classification performance. Future improvements could include optimizing kernel functions and extending the concept to other classification problems.
研究不足
The approach is limited by the two-dimensional evaluation of three-dimensional data, which can lead to misclassifications. The system requires an initial stabilization phase and may incorrectly classify dynamic objects extended in x-direction.
1：Experimental Design and Method Selection:
The approach involves building an occupancy grid representation of the environment and correcting it with a tailored transferable belief model.
2：Sample Selection and Data Sources:
Real-world Valeo Scala LIDAR data and an IPG Carmaker simulation are used.
3：List of Experimental Equipment and Materials:
Valeo Scala LIDAR sensor with specific parameters.
4：Experimental Procedures and Operational Workflow:
The method includes generating a scan map from sensor data, correcting it with the transferable belief model, and classifying cells as stationary or dynamic.
5：Data Analysis Methods:
The algorithm's performance is evaluated qualitatively and quantitatively.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
