A Tiny Chip Could Make LiDAR Much More Powerful: Here's How
LiDAR systems represent one of the key elements for environmental perception by autonomous vehicles, thanks to their ability to measure distances using infrared light pulses and reconstruct high-resolution three-dimensional maps. Despite high performance, many of the currently deployed solutions continue to rely on moving mechanical components, which increase size, costs, and wear over time.
A team of researchers from the Massachusetts Institute of Technology (MIT) has proposed a possible alternative based on a silicon photonic chip, designed to overcome some of the major limitations of solid-state LiDAR. The study, published in the journal Nature Communications, describes a new architecture of integrated optical arrays (Optical Phased Array, OPA) capable of significantly expanding the field of view while maintaining high precision and a low level of noise. According to the team led by Jelena Notaros, an associate professor in the MIT Department of Electrical Engineering and Computer Science, the result addresses one of the fundamental problems that have so far limited the evolution of fully integrated chip-based LiDAR.
In traditional LiDAR systems, scanning the environment occurs through a rotating module that directs light pulses in multiple directions. Silicon photonic-based systems, on the other hand, eliminate moving parts and steer the light beam by modifying the phase of the light distributed to a series of antennas integrated into the chip. This solution makes the sensor more compact, robust, and potentially less expensive, but it introduces an important limitation: the field of view is restricted.
The problem arises from the arrangement of the antennas. If positioned too closely together, strong electromagnetic coupling (crosstalk) occurs, altering the behavior of the light and degrading the quality of the beam. To prevent this, most designs space the antennas further apart. However, this choice generates secondary light beams that can produce false readings, reduce energy efficiency, and limit the maximum scanning angle.
To overcome this compromise, the researchers developed a set of three antennas with different geometries. By varying the width, size, and arrangement of the corrugations present along each antenna, the team managed to obtain different propagation coefficients, drastically reducing interaction between adjacent elements while keeping them very close together.
However, this approach requires delicate design balance. Although the antennas have different structures, they must emit the same amount of light, generate a beam with an identical angle at the same wavelength, and uniformly modify that angle during scanning. To achieve this, the team first developed a theoretical model dedicated to the coupling of radiative modes and then used it to design and simulate the entire system. Prototypes were then built and experimentally tested. According to the published results, the coupling between antennas decreased from values nearing 100% observed in a traditional configuration to about 1%, allowing the generation of a well-defined single beam and precise scanning over a much wider field of view, without the appearance of secondary beams.
The research group aims to further extend the scanning field and is exploring new possibilities that emerged during the development of the theoretical model. If confirmed by future developments, these results could pave the way for a new generation of solid-state LiDAR intended not only for autonomous driving but also for aerial surveying, construction site monitoring, and other applications requiring compact, reliable, and high-performance three-dimensional sensors.