Ateam of MIT researchers has developed a new imaging technique that could allow quality control robots in warehouses to peer into closed boxes. Using this new technology, robots could peer into a cardboard shipping box and see that the handle of a mug is broken, for example. The new method could revolutionize warehouse quality control and streamline the shipping and delivery process. The MIT team’s new approach utilizes millimeter wave (mmWave) signals to create accurate 3D reconstructions of objects. These are the same type of signals used in Wi-Fi, and they enable reconstructions, even if an object is blocked from view. These waves can travel through sheets of plastic, interior walls, and cardboard. They then reflect off hidden objects, allowing a sensor to collect these reflections. The MIT team’s system, called mmNorm, does just that. It feeds the wave reflection data into an algorithm, which estimates the shape of the object’s surface. During tests, mmNorm achieved 96 percent reconstruction accuracy. It detected and reconstructed objects like silverware and power drills. The research team specifically looked to reconstruct objects with complex structures and curvy shapes. According to the research team, similar state-of-the-art systems achieve 78 percent accuracy today.

Full report : MIT researchers develop a technique called mmNorm which lets humanoid robots see through closed boxes and walls.