Traditional solid-state LiDAR systems quickly scan three-dimensional shapes but cannot identify the substance of the objects they detect. A newly developed system pairs a silicon photonic array with a secondary photonic chip to read the polarisation of returning light. This allows the system to label materials like wood, stone, and steel point-by-point, giving robots and vehicles more context about their surroundings.
Optical phased arrays let solid-state LiDAR scan quickly, but ordinary versions mainly recover an object’s shape. The study adds another photonic chip beside the silicon scanning array. That second chip reads the polarisation of returned light and assigns a material category to each point, alongside depth, reflected-light strength and a polarisation feature.[1]
In laboratory trials at distances no greater than 10 metres, the device distinguished wood, stone and steel with accuracy above 85 percent. Its photonic inference unit returned classifications within nanoseconds and required little power for that module. Against the electronic back ends normally used for polarisation classification, the optical core supplied the speed advantage.[1]
While the technology expands what a robot or vehicle can interpret from a scanned scene, the current demonstration remains constrained in its scope. The physical tests only covered three specific materials at a relatively short range. The system does not yet demonstrate the same classification behaviour across a wider set of surface types or in complex real-world environments like active traffic, indicating further development is necessary before broader deployment.[1]
