KEF Robotics develops software for visual autonomy in aircraft, utilizing AI-powered obstacle detection to enhance military surveillance capabilities. Their technology addresses the need for improved situational awareness and mobility in complex environments, enabling more effective operational responses.
Funding
Funding not disclosed
TRFounders
Product
Problem
Tethered drones offer persistent aerial surveillance, but their maneuverability is limited by the physical tether, creating a risk of collision with obstacles in complex environments. Existing obstacle avoidance systems are often expensive, require specialized hardware, or lack the range necessary for safe operation at high speeds.
Solution
KEF Robotics develops AI-powered visual autonomy software that enables tethered drones to navigate complex environments safely. Using cost-effective electro-optical and infrared cameras, their system creates a real-time map of the drone's surroundings, identifying potential hazards such as power lines and tree branches. Machine learning algorithms and stereo vision techniques allow the drone to probabilistically map these hazards, enabling autonomous obstacle avoidance. This solution enhances situational awareness and mobility for tethered drones operating in dynamic and GPS-denied environments, improving operational effectiveness and safety.
Target Audience
The primary target audience includes military organizations using tethered drones for surveillance and reconnaissance, as well as commercial entities requiring autonomous navigation for tethered drones in complex environments.
Features
- AI-powered obstacle detection using electro-optical and infrared cameras
- Real-time mapping of surroundings using machine learning and stereo vision
- Autonomous navigation in complex and GPS-denied environments
- Hazard detection at sufficient range for path planning and obstacle avoidance
- System adaptable to multiple vehicle platforms
- Passive system that does not emit radio or electromagnetic frequencies
- Falcon simulator implementation for training autonomy systems in various scenarios