Collision Avoidance for autonomous UAV based on AI and computer vision
Klepsydra Collision Avoidance is a fast-responsive and intelligent vision based on-board software application for UAVs to perform vision-based navigation with autonomous drones. This software product has the following features:
- Artificial intelligence based computer vision for navigation
- Real-time response.
- Adaptable to several types of sensors.
- Lightweight and can fit in CPU only on-board computers.
- Compatible with main UAV autopilots and software frameworks.
Klepsydra Collision Avoidance works standalone, offering a waypoint and trajectory control interface in order to perform the trajectory desired by the mission control.
Safety and reliability
Zero data losses. Klepsydra can guarantee that any sensor data or communications message is going to be processed.
Real-time processing. Klepsydra can process data in real time with much less latency than other solutions, like ROS.
Predictable applications. Klepsydra Collision Avoidance is more stable, predictable and deterministic than other edge solutions.
Less hardware cost. Klepsydra Collision Avoidance is lightweight and can be installed in virtually any onboard computer with or without GPU.
Flexibility and compatibility
Klepsydra Collision Avoidance can be used with a number of different sensors including:
- RGB cameras
- perception cameras,
- Infrared cameras
Klepsydra is compatible with several frameworks and UAV technologies including:
- MAVLink, PX4, MAVROS
- ROS, Gazebo
- Linux, OpenCV.