Lidar technologies use infrared light pulses to measure distances and create detailed three-dimensional maps of the surrounding environment. This enables autonomous vehicles to detect obstacles and quickly respond to changes in their surroundings. However, traditional lidar sensors are often large, expensive, and contain moving parts that wear out over time, limiting their use in various applications.

A New Approach to Lidar Sensor Design

Researchers at the Massachusetts Institute of Technology have developed a method that could lead to more compact and durable lidar sensors without moving components. At the core of this method is a new chip based on silicon photonics—a semiconductor device that manipulates light instead of electrical signals. Existing silicon photonic lidars typically have a narrow field of view, making it difficult to scan the edges of a scene. Previous attempts to widen the field of view resulted in increased noise and reduced measurement accuracy.

The MIT team addressed these challenges by creating an array of integrated antennas that significantly reduces unwanted crosstalk between neighboring antennas. This design allows the chip to scan a wider field of view with less noise compared to other silicon photonic methods.

How Silicon Photonic Lidar Works

Traditional lidar systems use a large rotating block to direct light pulses across a scene. The light reflected from objects returns to the sensor, enabling the creation of a detailed map of the environment.

Silicon photonic lidars work differently: instead of mechanical rotation, the light beam is scanned electronically in multiple directions using an integrated optical phased array (OPA). The OPA consists of a group of integrated antennas, each containing microscopic grooves that cause light to scatter upward and exit the photonic chip. The direction of the beam is controlled by changing the phase of the light supplied to each antenna, allowing precise beam steering without moving physical components.

Solving the Crosstalk Problem

Dense placement of antennas leads to strong interactions between neighboring elements, which distorts the emitted light. Typically, to prevent interference, the distance between antennas is increased, but this creates multiple beam copies at different angles (grating lobes), reducing the system's efficiency and accuracy.

The MIT researchers developed antennas with reduced crosstalk, allowing them to be placed closer together without significant interaction. Instead of identical antennas, they created a repeating set of three antennas with different shapes, widths, and groove arrangements. Thanks to these varied geometries, each antenna has its own propagation coefficient, determining how light passes through the structure. This enables the antennas to operate with minimal interaction even when densely packed.

Experimental Results

In a typical OPA, interaction between antennas can reach 100%. In the MIT design, this figure is reduced to just 1%, resulting in a single clean and precise beam. The system accurately controlled the beam over a wide field of view without generating grating lobes. This combination of wide scanning, low interference, and high beam quality addresses one of the key challenges in integrated lidar technology.

Development Prospects

This achievement could pave the way for more advanced lidar sensors for complex tasks, including autonomous vehicle navigation, aerial photography, and construction site monitoring. The researchers plan to further refine the method to expand the field of view even more and are exploring alternative approaches that emerged during the development of the theory.

The research is supported by various scientific foundations and organizations, and part of the work was carried out using MIT.nano equipment.