In a “crowded” ToF application scenario, how to avoid interference between multiple ToF measurement signals?

Compared with the other two 3D depth vision solutions, the advantages of the current ToF solution in practical applications make the market optimistic about the greater opportunities of ToF in the fast-growing channel. For example, when calculating the depth of field after the picture is taken, no post-processing is required, which can avoid delays and save the related costs of using a powerful post-processing system; ToF ranging has a large scale flexibility, and in most cases it can be completed by simply changing the light source intensity, optical field of view, and transmitter pulse frequency.

In addition, due to its multiple advantages such as being not easily disturbed by external light, small size, fast response speed and high recognition accuracy, ToF is becoming the preferred technical solution for 3D vision in both mobile and automotive applications.

Using ToF technology can get the absolute value of the object, which has greater performance advantages

Autonomous obstacle avoidance for various robots in automated factories will be an area where ToF can be quickly applied. As robots are increasingly popular in industrial environments, when they are in a crowded working environment, they must be able to recognize humans and machines as well as the movements of machines, and respond quickly to avoid injuries to equipment and workers. If lidar is used to solve this problem, the cost will increase by tens of thousands of yuan, and the dual-camera solution requires a lot of calculations and precise positioning of the dual cameras. ToF has become the most cost-effective choice to solve the above problems.

ToF enables facial recognition to help upgrade buildings to smart. Taking the ADI ToF 3D stereo imaging automatic door solution with facial recognition as an example, traditional automatic doors use the infrared reflection principle and can only detect whether an object appears in the sensing range, which allows animals to enter and exit the mall freely, causing management problems. The ToF-based solution can identify human features in the space and the relative position and distance between people and objects to prevent non-humans (such as animals) from entering commercial stores. In addition, there have been mature solutions for automatic 3D crowd counting in commercial spaces in the past, but how to effectively use imaging technology to accurately distinguish the height and weight of people entering and leaving, while keeping the entry and exit time and height errors below 1%, this is a considerable technical threshold. ADI's ToF solution is different from traditional 3D crowd solutions, most of which require the installation of at least two stereo cameras. ADI's solution only requires a ToF camera lens holder, which is located above the door frame and has no installation height restrictions.

ToF applied to 3D crowd counting solution scenario virtual map

ToF car reversing images increase the car's driver assistance function. Combining image sensors and VGA ToF sensor modules with built-in image processors, ADI's automotive ToF solution can superimpose actual images and accurately measure the distance between objects and cars. Compared with traditional ultrasonic sensing solutions, it has a better sensing angle and can provide a wider range of collision detection and prevention for the reversing system.

The demonstration setup is very simple, with two ToF cameras set up perpendicular to each other, each camera will display the angle of the participating object in three dimensions, namely height, width and most importantly, depth. Each color represents a different distance range, so if the field worker gets closer to the camera, the color on the hand will change. Similarly, if you move away from the camera, the worker's body will turn a different color, which means the distance is getting farther.

This in itself is not a big deal, as there were many other 3D cameras on display at CES, but in this demonstration, ADI will show that multiple ToF cameras can be placed face to face without interfering with each other.

ToF cameras capture distance information by emitting laser pulses, then measuring the time it takes for the reflected pulses to return to the camera. By multiplying the reflection time by the speed of light, the desired distance or depth information can be obtained. The problem is that if there are other light sources with the same wavelength as the laser, especially other ToF cameras, the measurement will be disturbed. Therefore, the time measured in the presence of any type of light interference is incorrect, which means that the calculated distance information is also incorrect.

Comparison of the effects of multi-TOF interference cancellation technology demonstrated at CES2019 (used on the right, not used on the left)

You can see the effect of this interference from other cameras. Notice in this picture the color of the worker has changed, but this is meaningless because different colors represent different distances, this is due to the laser pulses from this ToF camera reflecting off the other ToF camera, thus showing distorted or wrong distance data.

On the other hand, this camera does not show this distortion because we use a patent-pending algorithm that avoids or eliminates all extraneous light information and only uses the light information from its own laser source, so it can give the correct depth information. The other camera does not use this interference elimination algorithm and will have distortion, but this camera will not have distortion.

Similarly, if a group of people wearing AR glasses are playing games in the same room, multiple autonomous robots are sorting goods in the same large warehouse, or two self-driving cars are approaching an intersection at the same time, the scope of applications using ToF technology for accurate depth measurement will be severely limited if the ToF camera cannot eliminate interference from other light sources. Therefore, ADI expects that the ability to prevent or eliminate interference in time-of-flight systems will become increasingly important.