Story perspectives

MIT Team Uses $100 LiDAR to See Around Corners

5/22/2026

Summary

Somasundaram Enables Corner Imaging

Siddharth Somasundaram’s MIT Media Lab team demonstrated that a smartphone-grade LiDAR can reconstruct hidden 3D objects and track motion around corners.
Using about 100 laser-emitter/detector pixels and motion-induced aperture sampling, they fused handheld frames to image a U-shaped object and bouncing ball on surfaces successfully.
The system cost under $100, needed no calibration, and included publicly released code and data for applications like warehouse robots, AR headsets, and autonomous vehicles.

Summary

Willomitzer Achieves Imaging

Since 2009, NLOS imaging research has used optical, thermal, acoustic, and holographic methods.
Kirmani et al. (2009) introduced transient imaging.
Velten et al. (2012) applied ultrafast time-of-flight imaging.
Katz et al. (2014) and Batarseh et al. (2018) created passive speckle-correlation and spatial-coherence techniques.
Willomitzer et al. (2021) achieved high-resolution, wide-field imaging via synthetic-wavelength holography.

Summary

NeRF Expands NLOS

Shen et al. (2021) and Fujimura et al. (2023) applied neural implicit representations to NLOS imaging.
Li et al. presented the transformer-based NLOST method at CVPR 2023, advancing deep-learning reconstruction.
Nam et al. achieved 5 fps low-latency time-of-flight imaging in 2021, while Pumarola et al. extended NeRF to moving objects that year.
Citations of NeRF (2021) and KinectFusion (2011) and publications in CVPR, ICCV, and SIGGRAPH show NLOS research merging with broader 3D reconstruction fields.

History

Historical Context

Microsoft launched the Kinect depth sensor on November 4 2010, selling 8 million units in its first 60 days and offering a 640×480 infrared camera at a low price. Researchers such as MIT Media Lab’s Siddharth Somasundaram leveraged this affordable hardware, building on earlier NLOS work by Kirmani (2009) and Velten (2012), to create a sub-$100 handheld 3-D imaging system now used in warehouse robots and AR headsets.

1 of 4

Somasundaram Enables Corner Imaging

Siddharth Somasundaram’s MIT Media Lab team demonstrated that a smartphone-grade LiDAR can reconstruct hidden 3D objects and track motion around corners.
Using about 100 laser-emitter/detector pixels and motion-induced aperture sampling, they fused handheld frames to image a U-shaped object and bouncing ball on surfaces successfully.
The system cost under $100, needed no calibration, and included publicly released code and data for applications like warehouse robots, AR headsets, and autonomous vehicles.

1 / 4