New chip set to update Lidar and accelerate autonomous systems

Lidar has become one of the most important sensing technologies of the modern era. From self-driving vehicles and delivery drones to infrastructure monitoring and advanced robotics, the ability to create accurate three-dimensional maps of the world is increasingly essential. Now, researchers at the Massachusetts Institute of Technology (MIT) have reported a breakthrough that could make lidar systems smaller, more reliable, and more capable. By developing a new silicon-photonics chip that dramatically reduces signal interference, the team has demonstrated a way to steer lidar beams across a much wider field of view without relying on moving parts, overcoming one of the technology’s longstanding challenges. The research was recently published in Nature Communications.

The advance could have important implications for Canada, where autonomous transportation, mining automation, drone surveying, and smart infrastructure projects are growing rapidly. Lidar, short for “light detection and ranging,” works by sending pulses of laser light into the environment and measuring how long it takes for reflected signals to return. These measurements are used to generate highly detailed three-dimensional maps of surroundings.

Unlike conventional cameras, lidar provides precise depth information, allowing machines to determine exactly how far away objects are. This capability has made the technology central to autonomous navigation systems. Modern self-driving vehicles often combine lidar with cameras, radar, and artificial intelligence to understand their environments. Drones use lidar for aerial mapping, while engineers employ the technology to survey construction sites and monitor critical infrastructure.

Many of today’s highest-performing lidar systems rely on mechanical components that physically rotate or move to scan an environment. While effective, these systems are often bulky, expensive, and vulnerable to wear over time. Researchers have long sought alternatives that eliminate moving parts altogether.

One promising solution involves silicon photonics, a field that uses semiconductor chips to manipulate light in much the same way conventional electronics manipulate electrical signals. Silicon-photonics-based lidar systems can be dramatically smaller and potentially less expensive to manufacture at scale. However, these chip-based systems have historically suffered from a restricted field of view. A narrow field of view limits how much of the surrounding environment a sensor can observe at any given time, creating obstacles for applications requiring comprehensive situational awareness.

A fundamental engineering challenge

At the centre of the problem lies a component known as an optical phased array (OPA). An OPA consists of numerous tiny integrated antennas that emit and steer light electronically. By adjusting the phase of light entering each antenna, engineers can alter the direction of the emitted beam without moving any mechanical components. To achieve a wide field of view, antennas must be positioned very close together. But when antennas are brought too close, they begin interacting with one another through a phenomenon known as crosstalk. This unwanted interaction distorts the emitted beam and degrades performance.

Historically, designers solved the problem by increasing the spacing between antennas. That approach reduces crosstalk but introduces another issue: unwanted duplicate beams known as grating lobes. These secondary beams reduce accuracy, waste energy, and limit steering performance.

The MIT team, led by Professor Jelena Notaros, approached the challenge from a different direction. Instead of building identical antennas, the researchers created a repeating set of three antenna types with different geometries. They modified antenna widths as well as the size and placement of microscopic corrugations that influence how light propagates through each structure. Because each antenna behaves slightly differently, neighbouring antennas interact far less strongly.

According to the researchers, this allows antennas to be positioned much closer together while substantially reducing crosstalk.The accomplishment is notable because all three antennas still had to emit light in a consistent manner. Achieving low interference while maintaining uniform beam characteristics required detailed theoretical modeling and extensive simulation work before fabrication could begin.

Under the test conditions, conventional optical phased arrays would typically exhibit coupling levels approaching 100 percent. The new MIT design reduced that figure to approximately 1 percent. At the same time, the system successfully generated a single clean beam that could be steered across a broad field of view without producing grating lobes.

In other words, the researchers were able to overcome the historical trade-off between wide-angle scanning and beam quality. According to Notaros, the work solves a fundamental problem that has restricted the performance of integrated optical phased arrays and could enable significantly more capable future lidar systems.

The MIT team plans to further expand the field of view and explore additional methods for improving chip-scale beam steering. Independent experts are already recognising the significance of the work. Joyce Poon, professor of electrical and computer engineering at the University of Toronto and director of the Max Planck Institute of Microstructure Physics, described the innovation as an important step forward for solid-state beam-steering technology. This Canadian connection highlights the broader international interest in the field.