In a joint effort between MIT CSAIL and Tianjin University, researchers have turned ordinary static objects into self‑aware tools that perceive and react to human motion. By embedding conductive filaments within the 3D‑printed material, sensors and wiring coexist in a single print.
Traditional single‑material prints suffer from signal interference, much like hearing two radio stations at once, which makes it hard for systems such as a robotic glove to distinguish the angle of a user’s movement.
The X‑Hinges software automatically inserts two distinct materials into a model: a filament that emits the strongest sensor signal and a separate trace material that does not interfere. This separation prevents cross‑talk even when multiple signal paths share the same object.
Users select shapes, connections, and up to three motion axes (compression, up‑down, side‑to‑side) through an intuitive interface. The tool places sensors at bend points—e.g., joints—so motion data from each axis is captured. Building a robotic hand, for instance, can record both vertical and lateral finger bends.
With X‑Hinges the team demonstrated several devices: an adjustable glove that records finger joint angles and streams them to a humanoid hand; an origami‑style lamp that senses a pinch and changes brightness; a platform that instantly recognizes a banana placed on it via tactile data; and a cartoon‑shark‑shaped joystick controller.
Unlike earlier approaches that allowed only a single axis of movement, X‑Hinges supports three axes and records multi‑directional data simultaneously, giving printed objects richer interactive capabilities.
A current limitation is calibration: each print introduces slight variations in the layout of electrical elements, so two visually identical items may produce different sensor readings. The researchers addressed this with a computer‑vision‑based recalibration system and aim to embed automatic signal correction in future versions.
The broader vision is to grow actuation and sensing together inside a single 3D print, creating new physical platforms for embodied intelligence and AI agents. The work will be presented at the ACM Symposium on User Interface Software and Technology (UIST) in early November.
Review