MIT LiDAR Breakthrough: Everyday Sensors Now See around Corners

by Aaron Leong — Thursday, May 21, 2026, 09:45 AM EDT

Massachusetts Institute of Technology (MIT) researchers are pushing consumer LiDAR from a depth sensor to a way to peek behind obstacles, which could be a huge boon for robotics, logistics, and even gaming.

The new work published in the Nature journal argues that a smartphone-grade LiDAR can "see" hidden objects that are reconstructed from the faint light that bounces off nearby surfaces and returned to the sensor. The team’s motion-induced aperture sampling model unifies object shape, object motion, and camera motion in one framework, which lets the system extract useful signals from scenes that would otherwise look like noise.

Years ago, MIT’s corner-seeing camera had to rely on lasers and picosecond timing to reconstruct geometry from scattered light, proving that indirect light could be decoded into a usable image. The newer MIT Media Lab approach described this time is much more practical: it uses everyday smartphone LiDAR hardware and computational tricks to achieve around-the-corner imaging without specialized calibration. In other words, the idea is now portable.

corner screenshot — *Real-time tracking of a hidden object*

With consumer LiDAR quietly becoming more common in phones, tablets, robots, and cars (where it already mostly helps with mapping, depth sensing, and navigation), this latest research suggests those same sensors can achieve non-line-of-sight imaging without bulky, specialized equipment. With the right motion and software, they can infer what is partially or fully hidden, which could improve robotics, assistive devices, inspection systems, and search-and-rescue tools.

Imagine having AR/VR headsets track your arms and legs when they're out of view, imaging systems that assist rescue teams search through obstructed spaces, or robots that can infer people or objects as they navigate cluttered spaces—the new possibilities and consumer experiences are almost endless.

*The reconstruction of the hidden object is still low-res, but it's a start*

Of course, there's still a gap between a lab demo and final end product. Hidden-scene imaging remains limited by low signal, sparse resolution, and the messy physics of light bouncing through real environments, and the technique will need careful testing across more scenes, materials, and motion patterns. Even so, the line between ordinary depth sensing and true around-the-corner vision is getting narrower. It also raises a new question about what counts as a “depth sensor,” at least in this case.

Image credits: MIT Media Lab