AI upgrade lets Qysea underwater drones visually lock onto targets
When remotely controlling an underwater drone in flowing water, it can be difficult to keep the vehicle's camera trained on a specific target. A new AI-based vision platform, however, allows Qysea Technology's existing FiFish drones to do so automatically.
Announced this Tuesday (June 7th), the system is known as AI Vision Lock.
It's activated via an updated version of the control app, which displays a live feed from the drone's HD camera on the touchscreen of a smartphone or tablet. Users simply tap the image of the stationary target – be it a coral head, bridge piling, damaged fish farm net or something else – on that screen. The drone responds by keeping that item centered in its shot until otherwise notified.
Utilizing a dead reckoning internal navigation system, the drone monitors its orientation and position in three-dimensional space, based on feedback from its accelerometer and gyroscope. If it detects that currents or other factors are pushing it off-target, it selectively activates one or more of its six thrusters in order to compensate.
Operators can also use the app to keep the target located off to one side within the camera shot, plus they can get the drone to move in on the target or move back, by making a finger-and-thumb pinch in/out gesture on the screen. Other aspects of the AI Vision Lock system are demonstrated in the video below.
It should also be noted that Notilo Plus' Seasam and iBubble underwater drones already offer similar functionality, to the point that they can actually move alongside mobile targets such as scuba divers.
Source: Qysea Technology