Can you provide a my_detection.launch file with complete perception features?

edit

@kosuke_murakami Thank you for your reply!

Can you check if this solution is correct（mainly for perception and local planning, ignoring mapping, localization, global_planning, etc)：

1. I use the vision_ssd_detect to detect objects in the vision input, and lidar_euclidean_cluster_detect to detect objects in the lidar input.
2. I use the range_fusion for sensor fusion.
3. I use the imm_ukf_pda_track to track the detected objects output by the fusion algorithm.
4. I use the naive_motion_predict to add motion prediction data.
5. I use the costmap_generator to generate costmap using the previous perception data.
6. Combining costmap, I use the astar_avoid and the velocity_set to generate waypoints
7. Finally, I use the pure_pursuit to generate vehicle control commands.