New Depth Camera
Hi everyone,
I’m pretty new to depth cameras and computer vision, but I just got my hands on a RealSense D415 (model 82635ASRCDVKhV) and I’m excited to start experimenting with it.
Here’s what I’m planning first: mount the D415 on a small wheeled robot platform. I want to stream the depth data to a laptop and visualize a point cloud in real time. After that works, I’d like to experiment with simple obstacle avoidance or “room mapping.” I’ll use the official RealSense SDK 2.0, seems like the most beginner‑friendly route.
Because I’m new to this, I have a few questions (and would love input from anyone who’s done similar builds):
-
What frame rate and depth resolution do you recommend for a moving robot? I want a stable feed but I also care about responsiveness, what’s a good balance?
-
Have you mounted the D415 on a robot or mobile setup before? How did you handle vibration, cable strain, and power supply stability?
-
In indoor lighting (with mixed light and shadows), did you notice issues with depth “noise” or flicker, and how did you mitigate them?
If any of you have run a D415 or other D400-series camera on similar projects, I’d appreciate your real-world insights.
Thanks in advance!
-
Hi Mercio Jonova Thanks very much for your questions.
1. 1280x720 resolution is the depth resolution that will provide optimal accuracy on the D415 camera model. Using the lower 848x480 should be fine too though if you find that you need to reduce the processing burden. In regard to FPS, 30 should be fine, and in fact 1280x720 resolution does not support a higher FPS than 30.
If the robot will be fast-moving then using 848x480 and 60 FPS may provide better stability because the images will be updating more frequently.
D415 is not best suited for fast-moving mobile applications (faster than a quick-waving human hand) because its depth and RGB sensors both use a slow Rolling shutter instead of the faster Global shutter. Using 60 FPS will help to minimize lag / blur on the images when moving quickly or observing fast motion.
2. The RealSense depth sensing algorithm is robust against vibration, but there may be problems if the vibration is severe. Introducing a vibration damping pad on the mounting point of the camera may help.
If you use a USB cable with screw-locks on the end then you can firmly mount the micro-sized end of the cable to the side of the camera using screw-thread holes either side of the USB port on the camera to restrict the connector from moving within the port.
The camera will be drawing power from the USB port on the robot, which in turn will be drawing power from the power supply unit of the robot's onboard computer. So power stability to the camera will be dependent on the power delivery stability of the USB ports of the particular computer onboard the robot.
3. If an area has both light and shadow, a feature called High-Dynamic Range (HDR) can be used instead of auto-exposure to define an exposure range with two separate values.
https://dev.realsenseai.com/docs/high-dynamic-range-with-stereoscopic-depth-cameras
Please sign in to leave a comment.
Comments
1 comment