3D-Sensing

Stereoscopic vision and sophisticated data processing in the brain enables humans to sense their environment in three dimensions. An increasing number of emerging technologies also demands this 3D-Sensing capability from machines and systems, often with stringent requirements for the safety and reliability of each method.

A popular and proven approach to 3D-Sensing, the 3-dimensional detection of the environment, is distance determination by measuring the time of flight of light, the time-of-flight (ToF) method. Here, a laser pulse is emitted, reflected by an object and then detected on an image sensor. The distance of the object can be determined from the duration of the time of flight and the speed of light. As LiDAR (from Light Detection and Ranging), this method is already frequently used, for example, in driver assistance systems, smartphones and robotics. More about this application in the field of "autonomous driving" can be found here.

The advanced development of lasers and the high speed of light allow in principle very fast distance determination and thus 3D-Sensing in real time. However, this places very high demands on the image sensor used, which must have the ability to reliably detect photons while determining the time of arrival with high resolution. In addition, if the distances of different points in space are to be determined, multiple pixels with independent time-determining elements are required on the image sensor.

Especially when LiDAR is used in autonomous vehicles, there are additional challenges, such as the background light of the sun, which can vary greatly and must still be distinguished from the laser light, which pose additional challenges to the image sensor. Since the transmitter (laser) side is largely limited here by the eye safety requirement, the main focus is on optimizing the receiver (image sensor) and subsequent data processing.

Expertise

The CSPAD sensors developed by Fraunhofer IMS combine highly sensitive 3D sensing capabilities with secure algorithms for high-performance use in LiDAR applications, among others. The central goal is to achieve maximum range even at high sunlight intensity and low system costs at the same time. Apart from the typical application area LiDAR, CSPAD sensors can also be used in other fields requiring time-resolved detection of photons like spectroscopy, quantum imaging and random number generation, among others.

The CSPAD technology is based on the integration of SPADs (Single-Photon Avalanche Diodes) in a 0.35 µm CMOS process certified for the automotive industry and optimized for optoelectronic applications. This allows SPAD sensor elements and read-out electronics to be accommodated on the same chip.  In addition, backside-illuminated CSPAD sensors can be realized using 3D integration in wafer-to-wafer and chip-to-wafer bonding processes.

The CSPAD sensors developed at Fraunhofer IMS are distinguished in particular by the adaptive photon coincidence circuits integrated into the pixels for background light suppression. This is the only way to ensure an increase in range under high sunlight irradiation and reliable distance measurement even under changing weather conditions.

In addition to the development and production of new, also customer-specific, CSPAD sensors, the Fraunhofer IMS provides a wide range of services in the field of 3D sensing. Evaluation boards for CSPAD sensors as well as complete LiDAR cameras with corresponding software are offered, which allow the sensors to be evaluated in realistic scenarios. Furthermore, system simulations are carried out at Fraunhofer IMS and new methods for signal processing are constantly being researched in order to achieve an additional increase in range. The use of artificial intelligence for reliable detection and efficient noise suppression is also being investigated.

SPADEye2

The SPADEye2 line sensor offers extremely low-noise CMOS integrated SPADs in 2x192 pixels, which, combined with complex circuitry, allow time-resolved detection of photons. Each pixel contains an independent TDC (time-to-digital converter) for time detection and a circuit for counting incident photons. 4 SPADs in each pixel are connected by an adaptive photon coincidence circuitry for background light suppression. The SPADEye2 sensor can detect photons with a time resolution of 312.5 ps in a measurement window of 1.28 µs, theoretically covering a distance of 192 m at a resolution of about 5 cm.