Perception

[annb3]

I am using autoware with a custom CAD, vehicle and sensor kit. In the sensor kit, there are a RGBD frontal camera and a center-upper LiDAR. During the perception phase, setting it in camera_lidar_fusion working mode, if the obstacle goes away from the frontal position, out of the FOV of the camera, the perception is interrupted: both the detection and the tracking, and there are also some not detected elements in the back. I can imagine that the use of only one camera determines this condition.
However, is there a possibility to make it work only using the lidar in case no camera info is received (always in camera_lidar_fusion mode)? Differently, how many cameras should I have to obtain good results and accuracy in the perception phase?