New developments in 3D cameras


New developments in 3D cameras will help boost the use of robots

Written by Jørgen Læssøe


New developments in 3D cameras will boost the future use of robots. Applications such as bin picking demand the capcbility of producing reliable three dimensional data. A number of different methods are available to generate the 3D data, each with individual pros-cons.

Structured light

For many years structured light has been the most common approach to obtaining 3D image data. Typically done by projecting a laserline onto the surface of a moving object on a conveyor. Afterwards performing simple triangulation on the geometry of the camera, laser line and reflected laserline may extract the height information and 3D info. Companies such as SICK and LMI technologies have a well developed portfolio of structured light systems that come pre-calibrated. For many types of applications this is a solid and robust method. It does require either the object to be moving past the system or a movement of the system itself. A single system will also have limitations when it comes to occlusions on the object. 


Stereo imaging

Another commonly used 3D system is stereo imaging where two cameras are used to capture two separate images of a scene from two different positions. The location and optical parameters of each camera must be calibrated so that triangulation methods can be used to determine the correspondence between pixels in each image. Depth in the scene can be computed since the depth of each point is inversely proportional to the difference in the distance of the corresponding points and their camera centers. Positional and optical parameters of the two cameras must be accurately calibrated. An increasing number of pre-calibrated products such as from Point Grey are available and are often used to solve issues regarding safety, volume measurement and gripping objects with a robot.

An alternative to stereo imaging is motion stereo. In motion stereo systems, a single camera is typically mounted on a robot and takes the required images from two or more locations. Because the baseline between camera views generally can be longer than in pre-configured dual camera stereo systems, the disparity accuracy can be increased. However, such systems require the robotic positioning errors to be minimized since this will affect the accuracy of the 3D data.

Stereo imaging is now also available as linescan solution with very high accuracy. This type of solution is very suitable for industrial use in applications where precision and speed is required and e.g. structured laserlight is insuffiecient like metallic surfaces. Chromasens and Datalogic are just some of the companies with this type of product in the pipeline. 

A crossover between stereo and structured light, called projected textures is also available. This method projects known patterns such as crosses, circles, squares, dot-matrices, multiple lines and random dot matrices onto a surface, and applies stereo cameras to extract depth. The Ensenso from IDS is a recent development to use projected textures.


Time of flight

Time of flight products incorporate illumination sources and imagers that illuminate objects in a scene with a modulated light source and then measure the phase shift between the illumination and the reflection to determine distance measurement. The scene is either illuminated by a pulsed mode or continuous wave (CW) illumination source, with a corresponding method to detemine the phase shifts. Companies like Odos imaging and Basler already have products available and could find their way into the self-driven cars as just one of many sensors.


Other techniques

3D techniques that shows a lot of promise and will soon find their way to industrial use, includes shape from shading, fringe projection, plenoptic/lens array cameras.

What should you choose?

When chosing the type of 3D vision system to use, you first of all look at the object that needs measurement. For quality control applications of low-specular material a structured light system based on a laserline may be sufficient, while time of flight methods can be used to locate distant objects. Stereo imaging systems are better suited to imaging high-contrast objects. For highly specular objects, it may be more useful to employ projected texture techniques. Luckily, you can also consult JLI vision for advice.