How Visual Navigation Enables UAVs Operations in GNSS-Denied Environments
Interview with José María Pulido - Navigation Systems Lead at UAV Navigation-Grupo Oesía
What resolution and dynamic range is needed for the sensor? how does the image processing compensate for fluctuations in the image? what range does this give for the navigation system and what altitude does this relate to?
The Visual Navigation System (VNS) is designed to operate with standard, mission appropriate imaging sensors rather than relying on extremely high resolution or specialized optics. The key requirement is not raw pixel count, but image consistency, sufficient texture, and stable contrast to support robust feature extraction across changing environments.
To address natural fluctuations in illumination, contrast, and scene appearance, the VNS relies on advanced image processing and AI based algorithms. These algorithms dynamically compensate for variations caused by lighting changes, shadows, motion blur, or terrain appearance by normalizing image data and focusing on persistent visual features rather than absolute brightness values. This allows the navigation solution to remain stable across different times of day, weather conditions, and surface types.
From an operational standpoint, the effective navigation range is therefore not defined by a fixed distance or altitude, but by the availability of meaningful visual features within the camera’s field of view. In practical terms, this means the VNS supports navigation from low to medium altitudes, where sufficient ground texture is present, complementing the inertial navigation system by continuously correcting drift rather than acting as a standalone positioning sensor.
By design, the VNS works as part of a sensor fusion architecture within the VECTOR autopilot, combining visual information with inertial data. This approach ensures that navigation performance gracefully adapts to changes in altitude and operating conditions, maintaining reliable position estimation in GNSS denied environments without imposing restrictive altitude envelopes.
Is there a difference in performance between global and rolling shutter?
Yes, there is a difference. For the Visual Navigation System we exclusively use global shutter sensors, as they avoid image distortion caused by platform motion and vibrations. Unlike rolling shutter, global shutter captures the entire image simultaneously, which is essential to preserve geometric consistency and ensure reliable feature extraction for visual navigation in dynamic flight conditions.
What other factors have to be considered for the image sensor and how that fits into the autopilot?
Beyond basic image quality characteristics such as resolution and dynamic range, several system level factors must be considered when selecting and integrating an image sensor for visual navigation. These include latency, synchronization, robustness to vibration and motion, power consumption, environmental tolerance, and long term reliability under operational conditions.
The Visual Navigation System has been computationally optimized to process image data efficiently while maintaining deterministic control loops. Rather than storing or processing full images continuously, the system extracts and retains only relevant points of interest, which are used for navigation and mapping purposes. This approach significantly reduces data volume, memory usage, and computational load.
Equally important is how the image sensor fits into the overall avionics architecture, rather than being treated as a standalone component. In the case of the VNS, the sensor is tightly integrated with the VECTOR autopilot, which provides precise timing, synchronization with inertial measurements, and deterministic data handling. This ensures that visual information can be accurately fused with inertial data in real time, which is critical for stable navigation performance.
By designing the VNS as part of a fully integrated GNC ecosystem, UAV Navigation-Grupo Oesía ensures that image based navigation is not limited by sensor characteristics alone, but benefits from a robust autopilot architecture that maximizes reliability, scalability, and mission adaptability.
What are the preferred interfaces for the sensor and are these determined by the customer?
The Visual Navigation System (VNS) has been designed with integration flexibility as a core principle. Rather than enforcing a single interface, the system supports multiple standard communication options, such as CAN, serial interfaces RS-232, and Ethernet, to adapt to different platform architectures and integration philosophies.
The choice of interface is typically determined by the customer and the specific constraints of their air vehicle or mission system. This approach allows the VNS to be easily integrated both with UAV Navigation’s VECTOR autopilots, which offer extensive connectivity capabilities, and with third party avionics architectures.
Thanks to the high level of connectivity and scalability of the VECTOR autopilot family, customers benefit from a flexible framework that enables the integration of a wide range of peripherals, sensors, or mission specific payloads alongside the VNS. This flexibility is particularly valuable for platforms operating in complex or evolving mission profiles, where adaptability and ease of integration are key.
