OAK-D-CM4¶
Overview¶
OAK-D-CM4 is based off of the OAK-D-CM3 and it combines a host (RPi CM4) and the OAK-SoM to allow for a fully standalone and integrated solution for real-time Spatial AI.
The biggest difference between OAK-D-CM4 and OAK-D-CM3 architecture is that OAK-D-CM4 integrates a CM4 module, has gigabit ethernet, and is populated with MagJack supporting PoE. PoE can be used with additional PoE Hat. CM4 modules also have WiFi and Bluetooth connectivity.
Most OAK-D-CM4 devices have eMMC memory on-board. If you would like to flash a new image to it, follow the tutorial here. OAK-SoM is connected to the RPi CM4 via USB2 lines (on PCB).
Hardware specifications¶
Camera module specifications¶
Camera Specs |
Color camera |
Stereo pair |
---|---|---|
Sensor |
IMX378 (PY011 AF) |
OV9282 (PY010 FF) |
DFOV / HFOV / VFOV |
||
Resolution |
12MP (4056x3040) |
1MP (1280x800) |
Focus |
AF: 8cm - ∞ or FF: 50cm - ∞ |
FF: 19.6cm - ∞ |
Max Framerate |
60 FPS |
120 FPS |
F-number |
1.8 ±5% |
2.0 ±5% |
Lens size |
1/2.3 inch |
1/4 inch |
Effective Focal Length |
4.81mm |
2.35mm |
Pixel size |
1.55µm x 1.55µm |
3µm x 3µm |
RVC2 inside¶
This OAK device is built on top of the RVC2. Main features:
4 TOPS of processing power (1.4 TOPS for AI - RVC2 NN Performance)
Run any AI model, even custom-architectured/built ones - models need to be converted.
Encoding: H.264, H.265, MJPEG - 4K/30FPS, 1080P/60FPS
Computer vision: warp/dewarp, resize, crop via ImageManip node, edge detection, feature tracking. You can also run custom CV functions
Stereo depth perception with filtering, post-processing, RGB-depth alignment, and high configurability
Object tracking: 2D and 3D tracking with ObjectTracker node
Stereo depth perception¶
This OAK camera has a baseline of 9.0cm - the distance between the left and the right stereo camera. Minimal and maximal depth perception (MinZ and Max) depends on camera FOV, resolution, and baseline- more information here.
Ideal range: 85cm - 10m
MinZ: ~20cm (400P, extended), ~40cm (400P OR 800P, extended), ~80cm (800P)
MaxZ: ~17 meters with a variance of 10%
Extended means that StereoDepth node has Extended disparity mode enabled.
Power consumption¶
Most of the power is consumed by the RVC2 and by the Raspberry Pi Compute Module (CM), so the power consumption depends mostly on the workload of the VPU and CPU:
Base consumption + camera streaming: 2.5W - 3W
AI subsystem consumption: Up to 1W
Stereo depth pipeline subsystem: Up to 0.5W
Video Encoder subsystem: Up to 0.5W
Raspberry Pi CM: Between 0.5W and 10W, depending on the CPU usage
So the total power consumption can be up to ~15W if you are using all the features at 100% at the same time. To reduce the power consumption, you can reduce CPU computation on the RPi CM, or FPS of the whole pipeline - that way, subsystems won’t be utilized at 100% and will consume less power.
Operating temperature¶
The ambient operating temperature of RVC2 based devices is between -20°C and 50°C when fully utilizing the VPU.
Similarly to the Power consumption, max operating temperature depends on VPU utilization. The higher the VPU utilization, the more heat the VPU will generate. The RVC2 VPU can continuously operate at 105 °C, after which the depthai library will automatically shut down the device (to avoid chip damage).
To find out more, see our Operative temperature range documentation.
Limitations¶
If the OAK-D-CM4 device does not come with a built-in WiFi module, a Tenda Wi-Fi dongle is included in the box of the device as a workaround. To enable the dongle, you will need to install the necessary drivers. These drivers come pre-installed on the V10 (or higher) OAK-CM4-POE 64bit RPi OS image (available here).
Alternatively, you can the steps below (source) to install the drivers manually:
sudo apt-get install build-essential git dkms linux-headers-$(uname -r)
git clone https://github.com/McMCCRU/rtl8188gu.git
cd rtl8188gu
make
sudo make install
sudo apt install --reinstall linux-firmware
sudo reboot
After rebooting, the Tenda USB dongle should be operational.