MIT engineers have cracked one of the hardest problems in chip-based lidar: how to pack light-steering antennas close together without them jamming each other's signals. Their breakthrough, published Wednesday in Nature Communications, slashes antenna interference from nearly 100% down to just 1% — and could give self-driving cars a dramatically better view of the road.
The problem with lidar on a chip
Lidar — the laser-based sensing technology that lets autonomous vehicles 'see' the world in 3D — has traditionally relied on large, spinning mechanical units. Solid-state silicon-photonics versions exist, but they suffer from a narrow field of view because their antennas can't be placed close enough together without creating disruptive 'crosstalk' — signal interference between neighboring emitters.
When antennas are spaced widely to avoid crosstalk, the system produces multiple ghost beams at different angles, confusing the sensor and wasting energy. Previous chip-based lidars were effectively looking at the world through a narrow tube.
Three shapes, one solution
The MIT team, led by Professor Jelena Notaros of the Research Laboratory of Electronics, designed an array of antennas using three distinct shapes — each with different widths and light-scattering patterns. Because each antenna type has a different 'propagation coefficient,' neighboring antennas simply don't see each other, eliminating crosstalk while remaining densely packed.
"Because the antennas have very different propagation coefficients, when we put them close together, essentially each antenna doesn't 'see' the antenna next to it. Therefore, it won't couple with its neighbor," explained Andres Garcia Coleto, an MIT graduate student and co-author.
The challenge was ensuring that despite their different geometries, all three antenna types emitted light at the same angle and intensity. The team developed a theoretical framework from scratch, then validated it with computer simulations and physical fabrication.
From 100% interference to 1%
In experimental tests, a conventional optical phased array (OPA) would have produced coupling of about 100%. The MIT design reduced that to roughly 1% while steering a single clean, precise beam across a broad field of view with no ghost images.
"The functionality we demonstrated in this work solves a fundamental problem for integrated optical-phased-array technology, enabling future lidar sensors that can achieve significantly higher performance," Notaros said.
What this means for autonomous vehicles
A lidar sensor that is both compact and wide-angle — with no moving parts to wear out — could accelerate the adoption of self-driving technology in passenger cars, delivery robots, and autonomous construction equipment. The same chip could also be used for aerial mapping, drone navigation, and industrial site monitoring.
Joyce Poon, professor at the University of Toronto and director of the Max Planck Institute of Microstructure Physics, called the work "an important step forward for chip-scale, solid-state beam-steering technology."
Next steps
The team plans to refine the method for an even wider field of view. They are also exploring a secondary approach that emerged during their theoretical work — a serendipitous discovery that could push the technology even further.
The research was supported in part by the Semiconductor Research Corporation, the National Science Foundation, and MIT.nano facilities.
