4.4.2. Stereo Vision

👓 Stereo Vision Mimics human binocular vision by using two cameras placed at a known distance apart.

Stereo vision is a technological method that imitates how humans see the world with two eyes. Humans have two eyes located a certain distance apart, which allows them to perceive depth. Similarly, stereo vision uses two cameras set apart by a specific distance to capture two images of the same scene. This setup helps in depth perception.

Think of stereo vision like your own eyes: when looking at an object, each eye sees it from a slightly different angle. Your brain processes these two images and determines how far away the object is. Just like that, robots use two cameras to gather images and calculate distances.

● Calculates depth using the disparity between left and right images.

In stereo vision, depth is calculated by analyzing the disparity between the two images captured by the cameras. Disparity refers to the difference in the position of an object in the two images. When the distance between the cameras is known, the disparity can be used to calculate how far away an object is from the cameras, thereby providing depth information.

Imagine looking at a nearby tree with both eyes closed and just one open. You might find it challenging to judge how far away it is. However, when both eyes are open, you can accurately tell the distance. Stereo vision works similarly by comparing how an object appears in two different views to assess its distance.

● Provides dense depth maps useful for navigation and manipulation. ● Requires calibration for accurate results.

Once the depth is determined, stereo vision produces depth maps, which are detailed representations of the distances of objects in the scene. These depth maps are essential for navigation where a robot needs to judge distances to avoid obstacles or when manipulating objects to ensure precision. Calibration of the cameras is also crucial, as it must be precise for accurate depth measurements.

Consider a car with parking sensors that uses stereo vision to detect nearby obstacles. It creates a depth map that helps the car navigate into a parking space safely by understanding how close it is to other cars or walls. Calibration is like adjusting your glasses to ensure you see clearly; if the cameras aren’t set up right, the depth perception will be off.

● Depth cameras (like Intel RealSense) simplify 3D perception by combining color and depth sensing.

Depth cameras, such as Intel RealSense, enhance stereo vision capabilities by integrating color and depth imaging into a single device. These cameras provide both a color image and a depth map simultaneously, making it easier for robots to understand their environment without needing a complex setup of multiple cameras. This technology simplifies the process of creating 3D models and enhances interactive applications.

Using a depth camera is like having a special pair of glasses that lets you see not only what something looks like but also how far away it is. For example, a gamer using a depth camera can interact with a virtual object: they can see it, understand how far away it is, and move it with their hands as if it were real.

--