Team SinNRG: How mmNorm Reinvents 3D Vision

Have you ever wondered why we can’t just "see" inside a sealed package without opening it? While X-rays exist, they are expensive and carry radiation risks. Traditional tech like LiDAR and cameras are easily defeated by a simple sheet of cardboard.

Enter Team SinNRG and their project, mmNorm—a breakthrough in non-line-of-sight (NLOS) imaging that transforms how we inspect the hidden world.

The Problem with Traditional Vision Most modern imaging systems have a "blind spot" when it comes to opaque materials.

Cameras & LiDAR: Completely blocked by wood, plastic, or walls.

Existing mmWave: While they can penetrate surfaces, they often suffer from poor depth resolution (~4 cm), resulting in blurry, "ghost-like" 3D reconstructions.

The mmNorm Innovation: Mirrors, Not Just Maps Instead of just mapping where an object is (occupancy-based imaging), mmNorm treats 77 GHz millimeter-wave signals like light hitting a mirror. Most everyday materials—like the cardboard of a shipping box—are semi-transparent at this frequency.

By leveraging specular (mirror-like) reflections, the team can estimate an object's Surface Normal Field. Essentially, the system calculates the orientation of every tiny point on the hidden object's surface to build a highly accurate 3D model.

Key Technical Milestones

High Fidelity: Achieves an impressive ~96% F-score in reconstruction accuracy.

Speed: The entire process—from mechanical scanning to GPU-accelerated 3D reconstruction—takes only about 8.5 seconds.

Smart Materials: The scanning platform uses advanced materials like PEEK and PTFE-Epoxy composites because they are "invisible" to radar, ensuring the machine doesn't interfere with its own sensors.

How It Works: The 5-Step Pipeline The "magic" happens through a sophisticated software-hardware synergy:

Scanning: Three motorized pods move in a coordinated 360 ∘ dance around the target.

Preprocessing: Raw data is converted from radio waves into a "complex cube" of range, velocity, and angle information.

RSDF Estimation: The system uses a Relative Signed Distance Function (RSDF) to handle the mathematical ambiguity of surfaces.

Poisson Reconstruction: This "refines" the 3D field, balancing fine details with global stability.

Marching Cubes: A final algorithm "stitches" the data into a standard 3D triangle mesh that can be viewed in tools like Blender.

Real-World Impact This isn't just a lab experiment. Team SinNRG sees immediate applications in railway warehouses and logistics. Imagine a system that can scan thousands of sealed packages to flag contraband, hazardous materials, or damaged goods without ever slowing down the conveyor belt.

By bridging the gap between advanced physics and the factory floor, mmNorm is turning the "unseen" into actionable 3D vision.