It's compact. It runs the Robot Operating System 2 (ROS 2) and utilizes Luxonis' DepthAI to see its surroundings. It also can chase you around playing hide-and-seek.
Luxonis' rae is small robot with lots of cameras (see figure) that uses 3D imaging for simultaneous location and mapping (SLAM). It can be tied into the cloud-based RobotHub for development. I talked with Conor Simmons, Machine Learning Engineer at Luxonis, about rae and its capabilities.
The compact rea robot (120 × 120 × 32 mm) uses the Robotics Vision Core 3 that's also found inside the OAK Series 3 cameras. The open-source platform incorporates a 4K Sony MX214 image sensor on the front and dual OV9782 stereo sensors on the front and back. These provide 3D object tracking and object localization support. The robot comes with preinstalled software that works in conjunction with a smartphone application that can be linked via Bluetooth or Wi-Fi.
The programmable system comes with 4 GB of DRAM and 32 GB of flash memory for storage. There are six microphones and a 1-W speaker for communication. The robot is powered by a 5,000-mAh LiPo battery. It was developed as part of a Kickstarter campaign.