During data collection sessions for autonomous driving, a large amount of Lidar data is recorded in the form of point clouds at different time steps. Users want to browse and navigate through scenes according to the content displayed, for example, to navigate to a timestep when a lady stays at the middle of the road or to the timestep when a car just arrives at a cross road. A prior art method for solving this problem is to display the point cloud in a view and create a timeline slider near the point cloud view. This timeline slider displays the indexes of the limesteps. When users click on one of the index, the corresponding frame of the point cloud are depicted in the point cloud view. This way they can locate the interesting scenario by examining the time steps one by one. The prior art method is a common practice for displaying time dependent point clouds. In this practice, a desired scenario (timestep) is located within a long 3D log using a timeline slider. This is time consuming and unintuitive because users have to click through the many time steps to verify if the displayed timestep contains the scene they are interested in.
展开▼
