This post is also available in:
עברית (Hebrew)
Smartphone LiDAR sensors have become common in high-end devices, mainly supporting augmented reality features and depth sensing for photography. But despite their growing capabilities, these sensors still face a basic limitation: they can only map objects within direct line of sight. Anything hidden behind a wall, obstacle, or partition remains invisible to the device.
Researchers have now developed a new processing method that pushes consumer LiDAR beyond that boundary. Using a standard smartphone-style sensor and a specialized algorithm, the system can detect and reconstruct objects hidden around corners by analyzing tiny light reflections that would normally be discarded as noise.
Conventional LiDAR works by emitting pulses of light and measuring how long they take to bounce back from visible surfaces. The new approach focuses on faint secondary reflections, as in light that scatters off floors, walls, and surrounding surfaces after interacting with hidden objects. While these reflections are extremely weak, they still contain information about the concealed scene.
The algorithm collects this scattered light data across multiple frames while either the sensor or the hidden object moves slightly. It then combines information from different viewing angles to estimate the shape, location, and movement of objects outside the sensor’s direct field of view. Instead of requiring expensive laboratory hardware, the system operates using consumer-grade LiDAR components costing under $100.
During testing, researchers placed objects such as mannequins, cardboard cutouts, and letter-shaped targets behind partitions where the sensor could not see them directly. According to TechXplore, by pointing the LiDAR toward nearby floors or walls, the algorithm successfully tracked movement and generated rough 3D reconstructions of the hidden objects in real time.
At the moment, the system performs best when it has some prior approximation of the target’s shape. Future work is expected to improve performance with unknown or dynamically changing objects.
From a defense and security perspective, non-line-of-sight sensing could eventually support robotics, search-and-rescue operations, autonomous navigation, and surveillance in obstructed environments. Systems capable of detecting movement beyond direct visibility may be especially valuable in urban terrain or confined spaces where conventional sensors have limited coverage.
The research also highlights how computational processing is increasingly expanding the capabilities of existing consumer hardware far beyond its original design purpose.
The research was published here.
