Breakthrough With Solid-State Lidar Researchers Eliminate Optical Crosstalk in Lidar Chips

Related Vendors

FAULHABER Drive Systems

SIMTOS (Seoul International Machine Tool Show)

Previous lidar systems at chip level often failed due to a physical compromise: a wide field of view came at the price of high optical crosstalk, which ruined the signal quality. Researchers at MIT have now developed an optical phased array (OPA) that elegantly avoids this problem by using asymmetrical antenna geometries.

Mechanical lidar systems with rotating mirrors are precise, but error-prone, large and expensive. Automotive manufacturers and industrial robotics are looking for alternatives to a solid-state system. One promising approach is silicon photonics, or more precisely optical phased arrays (OPAs). These control light beams purely electronically via phase shifting, without any moving parts. However, developers have so far come up against a hard physical limit when it comes to scaling.

Field of Vision Versus Crosstalk

In order to scan a wide field of view with an optical phased array, the tiny optical antennas on the chip must be placed extremely close together. If the antennas are placed too close together, unwanted optical coupling, known as crosstalk, occurs. The light jumps from one antenna to the neighboring one, which leads to noise and massive losses in beam quality.

However, if the distance between the antennas is increased to prevent crosstalk, side lobes (aliasing) occur. The lidar sensor then detects ghost images and loses its range.

Asymmetry As the Key to Decoupling

A team of researchers at MIT led by Jelena Notaros has now found a way out of this dilemma, as revealed in a recent publication. Instead of lining up identical antennas next to each other on a chip, the researchers used a design consisting of three different antenna geometries.

The developers deliberately varied the widths and corrugations of the neighboring antennas. The principle behind this is known from high-frequency technology: The structural asymmetry of the waveguides breaks the resonance between them. The light in one antenna no longer sees the neighboring antenna as an ideal propagation path.

The antennas can be packed extremely tightly for a wide field of view, while optical crosstalk is reduced to an absolute minimum. In simulations and initial prototypes, the team was able to demonstrate that the beam can be deflected precisely, with low noise and without mechanical assistance over wide angles.

What This Means in Practice

This progress is highly relevant for hardware developers and system integrators. If solid-state lidars based on silicon achieve the performance of mechanical systems, the unit costs of standard CMOS production processes will fall dramatically. At the same time, the form factor shrinks to chip size.

High-resolution lidar systems that were previously too expensive, too heavy or too fragile are now within reach:

• Autonomous driving: Seamless integration of high-resolution sensors into the vehicle body.
• Drones and UAVs: Weight reduction for longer flight times for topography or construction monitoring tasks.
• Industrial robotics: Robust 3D environment detection without wearing parts.

The MIT team is now working on further scaling up the system and preparing it for commercial use in real environments. It should only be a matter of time before these architectural concepts are incorporated into the first commercial photonic ICs (PICs). (heh)