MIT разработала lidar-чип без движущихся частей и с широким обзором для беспилотных авто

MIT (Massachusetts Institute of Technology) engineers presented on July 22, 2026 a lidar-on-chip design that delivers a wider and sharper field of view without a single moving part — thanks to antennas of different shapes placed tightly next to each other without mutual interference.

What's the problem with lidar on a chip

Lidar (light detection and ranging) measures distance by sending laser pulses and timing their return — this is how a three-dimensional map of space is built. The main difficulty in moving such a sensor onto a chip is the tightness on the die: the closer together the antennas that emit and receive light are, the more they distort their neighbors' signals. MIT engineers got around this by making neighboring antennas different shapes: due to their different geometry, they stop "picking up" and mixing up each other's radiation. According to MIT, in tests this approach sharply reduced mutual interference while preserving a single narrow and precise beam.

Classic mechanical lidars solve the same problem by brute force — rotating mirrors and motors that physically scan space. Such scanners are bulky, expensive, and wear out over time. Moving all the optics onto a chip removes the moving parts, but runs squarely into the problem of interference between closely spaced antennas — and that is exactly what MIT's new work addresses.

How the beam is steered without motors

MIT's chip steers the beam with electronics, not mechanics. The principle is close to a phased antenna array in radars: the signal phase at each antenna is adjusted so that the combined beam points in the needed direction, and it can be "shifted" across a sector without a single moving element. That's precisely why it's critical that neighboring antennas don't interfere with each other — otherwise the beam blurs and loses precision. Antennas of different shapes allow the emitters to be packed more tightly while still steering a single sharp beam across a wide field of view.

• Developer — MIT engineers
• Zero moving parts: the beam is steered by electronics, not mirrors
• Antennas of different shapes sit tightly together without mutual signal distortion
• In tests — a sharp reduction in interference while preserving a single precise beam
• Target application — sensors for self-driving cars

Antennas of different shapes can be placed close to each other without

"mixing" the neighbors' signals — that's how MIT engineers describe the essence of their development.

What this gives self-driving cars

A wider field of view at the same precision is a critical parameter for autonomous vehicles. Lidar builds a three-dimensional map around the car, and the wider the sector without blind spots, the sooner the system notices obstacles and the safer it makes decisions at speed. A narrow field of view forces manufacturers to install several sensors around the body's perimeter or add mechanics — and that means extra cost and additional points of failure.

A solid-state design without moving parts is also cheaper and more reliable. There's physically nothing in it to break, and mass-producing chips is fundamentally cheaper than assembling precision opto-mechanics. For automakers, this is a path to lidars that can be installed in mass-market models, not just expensive research prototypes.

What this means

MIT's work removes two limitations of solid-state lidars at once — narrow field of view and interference between antennas on the chip. If the approach scales to mass production, self-driving cars will get a wider, sharper, and cheaper "view" of the road without fragile mechanics. It's one more step toward reliable lidar ceasing to be the privilege of expensive research vehicles.