There are great needs in sensing a urban scene, which is dynamic and complex. To cope with such an environment, two sensing systems have been developed by our Lab. 

The first one is a ground-based sensing system, which generates trajectories of moving objects passing the scene using a laser scanner networks. My research based on this system includes: 

1. a trajectory processing method that selects and corrects mistakes in moving object tracking results automatically; 

2. a trajectory analysis method that model the scene using tracking results and detect abnormal trajectories in the scene.

 The second one is a vehicle-based sensing system, which uses laser scanners, cameras and GPS/IMU to generate a detailed copy of a dynamic scene at the moment. My research based on the system includes:

 1. an object extraction method that extracts defined types of objects from laser sensing data with an attention mechanism;

2. an attempt on fusing laser data and ladybug video, where spatial and color information are merged to provide a richer representation of scenes.

Yipu Zhao