Artificial vision has grown considerably along with the rise of UAVs. As the professional drones in all kinds of applications nowadays. Since its arrival on to the market, artificial vision technics have been used throughout the years and at present. Today, great advances in this technology are still expected in the medium term. For example, to create missions in which drones are informed of their environment and process this data and act autonomously on it.

The techniques have evolved considerably. And they continue to enhance thanks to the significant role of Deep learning. At this point, through the training and learning of the aerial platform. It can detect objects, people or animals in a short time and in a very precise way. In addition, artificial vision can help us to aware obstacles, automate operations and facilitate an automatic and safe flight.

Most requested applications of Artificial Vision in the field of UAVs

Artificial vision has multiple applications in the drone’s world. Some of the most demanded functionalities currently would be:

Improved stabilization

UAVs have purposes such as inspection, monitoring, mapping, precision agriculture, and civil engineering. The accuracy in estimation of the drone’s location is crucial to obtain maximum performance from these applications.

Artificial vision can help us to adjust the UAV placement estimation. This can be thanks to using the horizon or another fixed point as a reference. Today drones base their position and attitude calculation on GPS. This is in addition to other sensors such as the IMU or the magnetometer. The use of real-time image processing systems will allow us to reduce the dependence on GPS. It will also allow us to estimate the position under degraded GPS conditions.

Artificial vision in UAVs mainly uses Optical Flow algorithms. In this system, the GPS provides the drone with the global position to make the path according to the axes system. In another hand, whilst the machine vision algorithm provides the position’s data relative to a defined coordinate axis for the position and attitude adjustments.

Artificial Vision applications in UAVs

Object tracking in UAVs

UAVs predominate in military and industrial areas. Most demanded applications are such as aerial photography, monitoring some areas or structure inspection, etc. Most of these actions can be carried out using artificial vision integrated into UAVs. In particular, the localization of moving objects and target tracking is widely used. In order to detect and follow up, the drone will use the images captured by the camera. The drone will analyze the environment and use algorithms to classify characteristics.

Sense & Avoid System

As well as, certain areas automation, like object detection, and tracking, the same techniques are used to implement new systems to avoid obstacles such as buildings or other vehicles in flight. And therefore, these systems make possible to avoid collisions between UAVs too. For this reason, different forms of artificial vision seek to develop Sense & Avoid Algorithms. Veronte Autopilot integrates algorithms for the avoidance of obstacles. Therefore, only the necessary sensors are needed to make the detection and be able to enable autonomous functions like sense and avoid.

Embention has also added to the artificial vision trend. Recently we designed a new version of Veronte Gimbal Pro. This new gimbal has a totally renovated aesthetics, more compact, lighter and has superior mechanical robustness. In addition, the gimbal has an RGB camera with Full HD resolution, 30x optical zoom, and many more features to implement the capabilities already offered by onboard video processing in real time.