The dream of"> The dream of">

Coherent Vision can help machines see like humans


Ralf J. Münster, SiLC Technologies



// php echo do_shortcode (‘[responsivevoice_button voice=”US English Male” buttontext=”Listen to Post”]’)?>

The dream of autonomous robots that can help with household chores, drive us and replace factory workers has been around longer than the Jetsons series released in the 1960s. The reality is that many companies have promised the availability of self-driving cars. ‘by 2020, but we are still waiting. And the situation is not too different in the field of service robotics and industrial automation.

Why is it taking so long?

A key aspect is the way machines perceive their environment. While robotic vision has come a long way over the past decade, thanks in part to higher resolution cameras and newer 3D vision technologies, it is still far less efficient than humans at perceiving its surroundings.

In fact, current approaches could have it all wrong from the start. For example, we humans don’t scan our surroundings with multi-megapixel resolution, then meticulously try to find the outlines of objects and compare them frame by frame to conclude their movement paths.

Imagine sifting through a multi-megapixel image with a magnifying glass, marking slight variations in contrast and color as possible objects, then doing this at ten frames per second to track and predict the movement of objects. hypothetical.

Human perception works differently. Our eye is an extension of our brain, preprocessing large amounts of information. As a result, we have many more motion sensitive cells in our retina, which immediately lets us know if anything is moving in our field of view. Only then do we use our high-resolution cells in the macula to identify objects of interest, a much more efficient way of recognizing, tracking, and predicting objects.

Step into new 3D vision technology available now to help machines perceive their environment much more like humans. This is called a cohesive vision.

The technology sends out coherent laser light and captures much more than information about the intensity of the returned photons. It also captures tiny frequency shifts induced by object movements and provides information about the material and surfaces detected via changes in polarization.

Existing 3D sensing technologies include direct and indirect time-of-flight approaches, projected IR models (aka structured light), and triangulation techniques such as stereo vision. Not all of them offer instantaneous motion information and typically suffer from significant tradeoffs in range, eye safety, crosstalk immunity, and accuracy.

Continuous frequency modulated wave (FMCW) or coherent 3D detection

This diagram illustrates the principle of 3D FMCW detection. A low-power transmit beep (green) is reflected from an object. The frequency offset between the return chirp (blue) is proportional to the distance and speed of the object. A high and low chirp are used to resolve the two values, distance and speed.

Rather than relying on the detection of variations in light intensity, a common method of existing 3D detection techniques, the coherent 3D detection approach relies on the low-power frequency chirps of a highly coherent laser. This is also known as Continuous Frequency Modulated Wave (FMCW) technique, already used in advanced radar sensors.

Coherent photons can travel hundreds of feet, interacting and picking up characteristics of the target, then coming back while remaining in a coherent state where they can be mixed with some of the outgoing light for near lossless amplification.

The back mixing with the outgoing photons results in a beat frequency which is downconverted from the optical frequency, i.e. in the terahertz region, to the low gigahertz region and can be easily analyzed by them. electronic circuits available.

The distance of the measurement results in an optical frequency shift. If the measurement point also has radial velocity, the reflected chirp adds a Doppler frequency shift.

Using a high and low chirp allows coherent 3D sensors to instantly resolve both the range and speed of each pixel. This capability effectively extends 3D detection to 4D, i.e. simultaneous detection of x, y, z and speed vectors of an object.

The mixing of the returned photons with some of the outgoing laser light results in almost lossless optical amplification, allowing much higher sensitivity and detection accuracy. Due to the higher detection sensitivity of coherent systems, a laser power level of the order of one hundred milliwatts is usually sufficient to measure objects hundreds of meters away, allowing this technology to be integrated on a chip for use in mobile applications.

Linearly polarized photons can further change their state of polarization when interacting with targets, allowing the detection of material and surface features, such as windows or human skin.

FMCW advantages and 4D machine vision

Table 1 SiLC - Comparison of 3D machine vision approaches
Compare 3D machine vision approaches. (Source: SiLC Technologies)

Advances in technology and cost reduction have made 3D vision an essential technology in industrial manufacturing automation used to improve productivity, efficiency and quality. With many competing technologies available, the choice of technology is often decided by applications, ranging from general quality inspections, validations, audits and screening to safety and security.

FMCW promises to improve the performance vectors in multiple dimensions, allowing more precise scanning at longer distances while being safe to the eye and safe from outdoor lighting conditions or multi-system crosstalk. On top of that, it offers native 4D vision by providing speed information with every measurement.

Figure 2 SiLC - Polarization information can help provide information about materials and surfaces.
Polarization information can help provide information about materials and surfaces. (Image: SiLC Technologies)

What is the hold-up?

Why has it taken so long to generalize coherent 3D detection systems?

The main challenge in creating an FMCW solution has been the low cost, high volume manufacturing of high performance components. The coherent approach requires lasers with long coherence lengths (narrow line widths) and coherent light processing to extract additional information carried by the photons.

This requires very precise and low noise optical signal processing circuitry to form a coherent receiver. In addition, polarization plays a role here, because a coherent beat will only work for photons of the same polarization. The stability of the wavelength and the linearity of the laser source are essential during the measurement; otherwise, the signal to noise ratio may be greatly degraded.

Creating such a stable, robust and precisely defined optical system with discrete components is very difficult and expensive. In order to solve this problem, SiLC Technologies created a solution that integrates all the necessary optical functionality into a single silicon chip using semiconductor manufacturing processes used to fabricate electronic integrated circuits.

In other words, the same approach behind very complex silicon integrated electronic circuits which already enable very low cost consumer products, can now be deployed to realize very complex optical circuits for photonic applications.

Figure 3 SiLC - Depth and Velocity Point Cloud
SiLC’s integrated silicon photonics 4D FMCW vision system provides depth and speed data for each pixel measured. On the left is a depth point cloud, followed by a camera image in the center and a speed point cloud on the right. (Image: SiLC Technologies)

The silicon photonics integration platform integrates high performance components into a single chip using mature semiconductor manufacturing processes, providing a low cost, compact and low power consumption solution. Silicon fabrication also offers affordable, high-volume scaling of complex devices and technologies.

Figure 4 SiLC - Eyeonic vision sensor
SiLC Eyeonic The vision sensor based on silicon photonic integrated circuits combines the critical components needed to realize an FMCW LiDAR transceiver in a small form factor.

In summary, 3D vision is essential for the perception of the machine. Coherent 3D sensing using the FMCW technique is the most recent of these technologies, extending the performance characteristics of vision systems to many levels, even in the fourth dimension. Rather than relying on time of flight, stereo vision, triangulation, or structured light, FMCW detection takes advantage of the properties of the photons themselves.

Exploitation of this approach was hampered by the cost and required number of components. The power of silicon integration and the legacy of bringing silicon photonics to market can be harnessed to finally bring a cost effective coherent vision sensor to market. By using additional instantaneous speed information, it will help machines perceive their surroundings more like humans.

– Ralf J. Muenster is Vice President of Business Development and Marketing for SiLC Technologies, Inc.


Comments are closed.