At MIT, researchers have developed mmNorm, a novel millimeter-wave (mmWave) imaging technique that enables accurate 3D reconstruction of objects hidden behind opaque materials such as wood, cardboard, or plastic. Unlike traditional mmWave systems that rely on low-resolution back-projection, mmNorm leverages specular reflections and uses surface normal estimation to reconstruct detailed object geometry, even when the object is fully occluded.
A story on the MIT website tells how a reconstruction algorithm (inspired by computer graphics) is used to generate a cohesive 3D model of the hidden object. By mounting mmWave radar on a robotic arm, the system collects reflections from multiple viewpoints. Each reflection provides “votes” for possible surface orientations, and these are aggregated to calculate precise surface normals. The result is a high-fidelity 3D shape reconstruction with up to 96% accuracy, outperforming conventional methods.
This technology holds promise for robotic automation, non-invasive inspection, and security applications, allowing robots to identify tools inside drawers or verify contents of sealed packages without opening them. mmNorm operates across a range of materials and requires no special lighting, making it highly versatile for industrial and field use.
Future directions of this technique can include the following:
- Boosting resolution for even finer surface detail.
- Improving performance on low-reflectivity materials and the ability to handle thicker occlusions.
- Exploring handheld or mobile forms of mmNorm sensors (e.g., drones and wearable mm‑wave scanners).
With further refinement, mmNorm could revolutionize how machines perceive and interact with hidden environments—bringing human-like spatial understanding to autonomous systems.