In recent years, machine vision systems used in robots and drones have been considered by many to be insufficiently fast to respond to unpredictable movements in the surrounding world, despite their high performance. These devices often require significant resources or have high costs, limiting their application in everyday life. However, in a recent issue of Nature, researchers presented a 4D imaging sensor on a chip capable of creating 3D maps of terrain while simultaneously tracking the speed of moving objects.
Chip with Sensors
As part of their research, scientists developed a focal plane array (FPA) — a grid of 61,952 stationary pixels created on a single silicon chip. Each pixel functions as a miniature sensor that emits laser light towards the scene being studied and captures the returning signal.
To provide "vision" of the surrounding world, laser light is supplied to the chip from the outside. It passes through a network of optical switches that sequentially direct it to groups of pixels. Each pixel uses the FMCW LiDAR method to analyze the returning signal, allowing it to calculate the distance to the object and its speed. Unlike many traditional LiDAR systems, where one set of pixels emits light and another receives it, in the new system, all pixels can perform both functions simultaneously, making the device more compact.
Additionally, unlike standard sensors, this chip operates with a continuous laser beam instead of short light pulses. This advantage allows for the detection of even minor changes in the frequency of light waves, enabling the simultaneous calculation of distance and speed of moving objects.
Joint Measurement of Distance and Speed
The research team conducted tests of their chip under various conditions to determine its effectiveness at different distances and in motion. The sensor successfully scanned scenes and created three-dimensional digital maps consisting of thousands of data points, displaying interiors at distances from 6 to 11 meters.
To assess the range, the scientists aimed the chip at a building located 65 meters away, and it was able to detail even elements such as windows and balconies. A test was also conducted on a rotating disk, where the chip instantly determined the rotation speed.
“This is the first demonstration of a large-scale coherent matrix of photodetectors with integrated all necessary electronics on a single chip, which significantly reduces costs and accelerates the implementation of the technology,” noted lead author Remus Nicolaescu.
In the future, the team aims to improve the chip's characteristics, including increasing resolution and extending range. Once it moves beyond the laboratory, this development could prove useful not only for robots and drones but also for enhancing digital cameras, including those used in mobile phones.
The material "New chip allows robots to see in 4D, simultaneously tracking distance and speed" was published on the K-News website.
