Robots learned to " see " through cardboard thanks to MIT technology

MIT scientists have developed an innovative mmNorm visualization system that allows robots to see through cardboard boxes and other opaque obstacles with record-breaking accuracy. This technology is based on the use of millimeter waves (mmWave) — similar to those used in Wi-Fi — to create accurate three-dimensional models of hidden objects.

The mmNorm system captures wave reflections from objects located behind obstacles, such as plastic containers or interior walls, and processes them through an algorithm that can estimate surface normals-vectors perpendicular to the object's points. The unique approach takes into account the phenomenon of mirror reflection of waves, which significantly improves the accuracy of object reconstruction. In tests, the technology showed 96% accuracy, surpassing existing methods with their 78%.

The main advantage of mmNorm is that there is no need for additional bandwidth, which provides ample opportunities for use in various conditions. An example of an application is quality control in warehouses, where robots can distinguish between tools hidden in crates and accurately determine their shape for efficient capture.

To create the prototype, scientists used a robotic arm with a radar that moves around a hidden object and collects data on reflections at different points. The system combines this data, allowing you to build three-dimensional models even for several items in the same box. The technology works successfully with materials such as wood, metal, plastic, rubber and glass, except when objects are hidden behind metal or very thick barriers.

In the future, the researchers plan to improve mmNorm resolution, improve performance when working with less reflective surfaces, and expand the ability to visualize through denser obstacles. Potential applications include robotics, augmented reality, and security systems, opening up new horizons in automation and hidden object analysis.