rae¶
Buy it on Luxonis shop - Early Access
Overview¶
rae (short for “Robotics Access for Everyone”) is a small desktop robot developed for evaluation of the DepthAI ecosystem and rapid prototyping of robotics applications. It is designed to be a low-cost, easy-to-use, and extensible platform. It was brought to life by a successful Kickstarter campaign.
It’s built on top of the RVC3, which along with AI, CV, video encoding, and stereo depth features, also supports on-chip VIO and Sparse SLAM. rae is running ROS2 which handles path planning and navigation.
Specification overview¶
Uses RVC3 for the computation with 4GB RAM and 32GB eMMC storage
Back side sensors: 2x OV9782 wide FOV
IMU: BMI270 6-axis IMU
Wheels: 2x GM12F-N20VAV DC motors with Hall sensor wheel encoders
Connectivity: Either USB-C (on the side of the robot) or WiFi (AP)
Battery: 5000mAh Li-Ion, about 1 hour of runtime
Microphone array: 6x I2S MEMS microphones, 60Hz - 20kHz
Speaker: 1W, 100Hz - 20kHz
Display: RGB TFT, diagonal 0.96”, 80X160 pixels
Camera Specs |
Color |
Stereo pair / Color |
|---|---|---|
Sensor |
IMX214 (color, PY138) |
OV9782 (color, PY139) |
DFOV / HFOV / VFOV |
||
Rectified Depth FOV |
N/A |
|
Resolution |
13MP (4208x3120) |
1MP (1280x800) |
Focus |
FF: 60cm - ∞ |
FF: 18cm - ∞ |
Max Framerate |
60 FPS |
120 FPS |
F-number |
2.2 ±5% |
2 ±5% |
Lens size |
1/3 inch |
1/4 inch |
Effective Focal Length |
2.26mm |
1.69mm |
Pixel size |
1.55µm x 1.55µm |
3µm x 3µm |
RVC3 inside¶
This OAK device is built on top of the RVC3. Main features:
3.0 TOPS for AI with INT8 quantization support
Quad-core ARM A53 @ 1.5GHz, running Yocto Linux, acting as a host computer
Imaging: ISP, max 6 cameras, 500 MP/s HDR, 3A
Run any AI model, even custom architectured/built ones - models need to be converted.
Cloud platform - Robothub - connectivity out-of-the-box
On-device SLAM / VIO support
Encoding: H.264, H.265, MJPEG - 4K/75FPS, Decoding: 4K/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
Hardware specifications¶
Utilizes OAK-SoM MAX (OAK-SoM-Max-1 configuration)
Wireless card: Intel 9260 (via PCIe M.2)
Accessory USB-C (on top) works as USB3.0 host, so you can connect USB devices to it
USB-C charging port (on the side) is USB2, while Accessory USB-C (on top) is USB3
LED ring (around the bottom): 10 LEDs per side, in total 40 LEDs individually addressable via WS2812 protocol
Stereo depth perception¶
This OAK camera has a baseline of 7.5cm - 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: 40cm - 6m
MinZ: ~20cm (400P OR 800P, extended), ~37cm (800P)
MaxZ: ~10 meters with a variance of 10% (depth accuracy evaluation)
Extended means that StereoDepth node has Extended disparity mode enabled.
Getting started¶
Press the power button, which is located on the bottom side of the rae, to turn it on. Press the power button twice to shut it down.
Deploying apps via RobotHub¶
RobotHub is our cloud control platform, allowing users to connect easily to our devices. You can just scan the QR code (generated for you), and the rae will connect to our cloud platform through your WiFi. This will allow users to easily deploy pre-made apps, such as Follow-Me, Floor/3D Mapping, Sentry Mode, etc. with a click of a button. As RobotHub frontend is mobile-native, these apps (their frontend control part) will also work on mobile devices and will be open-source, so users will be able to build their solutions on top of them. See the documentation here on how to connect the rae to the RobotHub.
Direct connection¶
Besides deploying apps (docker containers with scripts inside) via RobotHub, one can also connect to the rae directly via SSH, and control the robot that way. This path is more suited for developers who don’t mind diving deep into the perception, navigation, and ROS logic. The best path to get started would be through with the rae-ros Github repository.
Default IP address is 192.168.197.55/28. To connect to the RVC3 via SSH, run:
ssh [email protected]
# No password is needed.
rae will create WiFi access point by default, with the following settings:
SSID: rae-<ID>
Password:
wifiwifi@
After connecting to its AP, rae’s IP is 192.168.11.1/24. You can connect to it and SSH into rae:
ssh [email protected]
# No password is needed.
Running NN on rae¶
Documentation here Dev Blobconverter here: https://dev-blobconverter.luxonis.com/
Upload files to RAE¶
# If you use MAC you may need to use -O option to enable file transfers with scp
scp <file> [email protected]:/<path>
System partition is Read-Only. /data and /home are Read/Write.
Firmware update¶
Log into RAE and run:
mender -install <link_to_firmware> ; reboot
You can check the version of the OS by running:
cat /etc/os-release
Reset¶
Press the power button for 8s for a hard shutdown. You can factory reset rae by holding the reset button (with a pin)
for 10s. Factory reset will remove everything from /data and all user changes made to /etc and /var.
3D Models¶
3D models - STEP/STL of the enclosure and the PCBA - can be found here