User manual

33
˗ Highly reflective, such as water
˗ Especially dark or bright
˗ Subject to frequent lighting changes
˗ Made of closely repeating patterns, such as tiles
˗ Sound-absorbent, such as thick carpeting
˗ In motion, such as roads with heavy traffic
 Keep the binocular camera lenses and sensors clean.
 To avoid interference with the Starpoint™ Positioning System, do not use 40
kHz ultrasonic devices, such as ultrasonic rangefinders, fault detectors,
cleaners or welding machines.
 Functionality Description
 Dynamic Tracking
Dynamic tracking uses deep learning algorithms to detect six types of subjects in
real time: pedestrian, cyclist, car, truck, boat, or animal. A real-time tracking
algorithm is used to automatically track the selected subject while avoiding
obstacles during flight. The function can track the subject in three modes. For more
information please refer to the APP Manual document.
 Viewpoint
Viewpoint allows the user to send the aircraft to a selected destination by touching
a point on the screen.
 Gesture Commands
The main camera uses deep learning algorithms to recognize and respond to three
gesture commands: outstretch your arms to set yourself as a target, raise both arms
to capture a photo, and raise one arm to start or stop recording.
 Accurate Landing
Accurate landing uses the aircraft’s bottom binocular vision system to save a series
of images indicating the aircraft's altitude and yaw during takeoff. During the go-
home and landing process, the aircraft matches the altitude and yaw to the images
taken during takeoff, and calculates how far off the aircraft is from its takeoff location.
The aircraft is accurately controlled based on VIO feedback, and lands in its takeoff
position.