The output 3D models can be used for intelligence gathering, planning and simulation including the extraction of measurements, field of view, accessibility, etc. The ability to measure objects appearing in aerial footage from UAVs increases the accessibility and usability of such information.
Camera Based Navigation
Maintain Navigation Capabilities in GPS-Denied Zones
The proposed project is set to provide a navigation solution for UAVs (or UGVs) where there is no available GPS signal. D-Vision solution will act as an autonomous system and provide more accurate navigation data than any existing non-GPS technology. Using only the camera and computer already available on the platform, it is a low cost software solution with minimal maintenance and simple to upgrade. As a result, UAVs equipped with the system would gain a better performance and a more precise positioning without a GPS receiver.
Navigation Problems in GPS-Denied Regions
Many military companies require UAV (or UGV) activities in regions where GPS signal is blocked or jammed by enemy’s EW misleading signals. Such severe conditions won’t allow GPS based navigation.
Another problem is that newly developed mini UAVs can carry pods with limited weight and size in order to allow more payloads and make the whole system more efficient on fuel. Furthermore, severe budget constrains require the navigation solution to be cheaper than currently available expensive GPS receivers (costing up to $10K per unit) that are too big and heavy for mini UAVs.
Independent Vision-Based Navigation System
D-Vision develops vision-based GPS-denied navigation system that features a video camera (typically already exists on UAV platforms), a communication unit and a computer for image processing located in the Ground Control Station. The captured video will be transferred to the computer by downlink for real-time calculations. Based on this data, the computer applies the technology to generate a 3D model of the world appearing on the transferred frames. New 3D output as well as navigation coordinates will be available 5 times per second (refresh rate of 5 Hz). This data will be used to determine the UAV’s spatial position and direction and will be fed to the UAV’s autopilot by the uplink (there is an option to install a tiny computer or FPGA for video processing on the UAV). In order to align local UAV coordinates with global coordinates the system will be fed with its initial global position. From there on, the system will autonomously calculate its advancements.
Aerial Photography Enhancement
The system can be fed with aerial photography or pre programmed known points (i.e. photo of a house with known coordinates) in order to fully annul drift throughout the whole flight period. When flying over them, the system would recognize reference points and would align itself accordingly. This task will be done with pattern recognition and object classification algorithms. These reference points would allow nullifying navigation drift errors over the flight.
Electric Wires Alerting
Electric Wires Anti-Strike Aircraft Safety Aid
Aerial vehicles such as UAVs, helicopters and light aircrafts fly in the proximity of electric wires, elevated cables, pylons and masts. Identifying these objects in time to prevent a collision is of great need, since a collision with such obstacles is a statistical certainty, while the outcome of such an accident is extremely expensive and fatal in most cases.
D-Vision technology allows identifying these obstacles and notifying the pilot about the threat. The system consists of a single digital camera mounted on the aircraft body connected to a standard computer and the D-Vision solution for image processing. The camera is pointed towards the aircraft flying direction and will scan the space for potential obstacles. When the system detects such an obstacle, it provides a warning notification.
The detection can be limited to a desired range to prevent interference with distinct unthreatening obstacles. Obstacle identification can be done with a night vision camera as well, allowing for night flights.
Anti collision system
Object recognition & detection system
Real time notifications
UAV Landing System
UAV Vision-Based Landing System
Ironically, humans can’t land airplanes without being able to see the runway, yet we expect UAVs to do so all the time, using sensors that are not ideally suited to the task.
Ability to land UAV without human intervention by just using computerized vision technologies can sometime prevent UAV from crashing. Various aerial operations call for a remote pilot to land the UAV manually through the use of stick and throttle control inputs transmitted wirelessly. Operators are reliant on the ability to view the runway via RF broadcast video. Any loss of communication to the ground, link or video quality can have a dire effect on the successful outcome of the landing. By providing what is essentially a vision-based landing sensor, D-Vision provides a way to enable repeatable and safe landings when no remote pilot is available due to loss of control signal or no available RF link at an unplanned landing zone.
The vision-based landing system consists of D-Vision proprietary software technology, existing on-board video camera and standard CPU for image processing, integrated with on-board aviation and navigation systems. The system constantly provides with UAV’s spatial position to the autopilot by using existing, standard on-board camera only. No additional ground-based systems or specialized navigation equipment are required.
For extremely weight and space sensitive aerial platforms there is an option to place the computer for image processing on a Ground Control Station and connect it to the communication unit for bi directional data streaming. The captured video will be transferred to the computer by downlink for real-time calculations. Based on this data, the computer applies the technology to generate a 3D model of the landing area appearing on the transferred frames. New 3D output as well as navigation coordinates will be available 5 times per second (refresh rate of 5 Hz). This data will be used to determine the UAV’s spatial position and direction and will be fed to the UAV’s autopilot by the uplink.
More than 13 years experience was invested in developing systems that allow for the determination of position and orientation of a camera in real-time, simply by tracking features in an image sequence or scene. UAV Visual Landing System is a natural extension of that work, helping safely and reliably guiding a UAV from a low altitude to touchdown.
UAV Landing Aboard Carrier
Retrieval by netting is a common UAV aboard landing method that utilizes landing hook to catch the landing cable. They receive guidance commands from the ship that steers the aircraft towards the braking cable with a very high precision, depending on a UAV physical size. If the UAV fails to receive terminal navigation commands in real-time and misses out the cable it ends up in the water!
Another optional add-on is vision based landing aid in the terminal approach of a braking cable. The system will analyze the video and track the braking cable visually to determine the UAV’s angular position and range to the cable. The UAV’s position will be updated in real time relatively to the braking cable in all 6 degrees of freedom, and will be fed to the UAV’s autopilot.
These solutions provide augmentation to existing systems, ensuring the aircraft always gets back down in the event of the feared loss of control link.
Motion detection from a moving platform is regarded as a key component in many smart surveillance systems. Upon request, D-Vision can develop tailor made solutions based on its proprietary technology. The application for person-vehicle classification is based on a video footage taken in real-time. Its key advantage is the ability to track moving objects with a mobile camera, allowing it to be installed on a ground or aerial vehicle.
Replace or backup your GPS receiver
Navigate using a standard video camera only
Independency in the air and on the ground
Advantages of software based solution
Excellent for weight & space sensitive applications
Low cost & efficient