Lidar has become one of the key technologies behind autonomous vehicles. By firing pulses of infrared light and measuring how long they take to return, it can build a detailed three-dimensional picture of the world around it, allowing vehicles to detect obstacles and respond in real time.
The technology works well, but it comes at a cost. Many lidar systems rely on rotating mirrors or spinning laser assemblies to scan their surroundings. They are effective, but also relatively bulky, expensive and packed with moving parts that inevitably wear over time.
Engineers have long seen silicon photonics as an attractive alternative. Instead of moving mirrors, these semiconductor chips steer light electronically, offering the prospect of smaller, cheaper and more robust lidar sensors. The catch is that they struggle to see as much of the world around them. Compared with conventional systems, their field of view is often much narrower.
Researchers at the Massachusetts Institute of Technology (MIT) have now developed a way of tackling that problem.
Advertisement
Their approach focuses on one of the chip’s smallest components: an array of microscopic optical antennas. Working together, these antennas steer a beam of light electronically across a scene without any moving parts.
Packing the antennas close together allows the beam to scan across a wider angle. Unfortunately, it also creates a new problem. The antennas begin to interfere with one another, a phenomenon known as crosstalk, which distorts the outgoing light and reduces the sensor’s accuracy. Moving them further apart avoids the interference, but introduces additional beams, called grating lobes, that limit the useful scanning range and can even create false detections.
Rather than using identical antennas throughout the array, the MIT researchers took a different approach. They designed a repeating pattern of three subtly different antennas, varying their width and the arrangement of tiny corrugations etched into each one. Although the antennas are physically different, they were engineered to emit light in exactly the same way while interacting far less with their neighbours.
Designing the array was a balancing act. Changing the antenna geometry reduced interference, but also risked changing the way each antenna emitted light. To overcome this, the researchers first developed an electromagnetic model describing how the antennas interact before using it to optimise the design and fabricate the chip.
In laboratory tests, the new array reduced antenna coupling from around 100% in a conventional design to approximately 1%. At the same time, it maintained a single, well-defined beam across a much wider field of view without generating the unwanted grating lobes that have limited previous silicon-photonics approaches.
Advertisement
The advance could help support a new generation of compact solid-state lidar systems for autonomous vehicles, aerial surveying and construction site monitoring, where reliable, maintenance-free sensors are increasingly in demand.
More broadly, the work illustrates how the future of lidar is becoming less about mechanical engineering and more about photonic engineering. As researchers continue to integrate optical functions onto semiconductor chips, sensing systems that once depended on motors, mirrors and moving assemblies are steadily being replaced by devices measured in millimetres rather than centimetres. That shift could make lidar smaller, more reliable and easier to manufacture, bringing the technology into a much wider range of engineering applications.