Imagine a robot that thinks — not with chips or wires, but with its own body. In a paper “Reprogrammable sequencing for physically intelligent underactuated robots,” researchers show how you can build mechanisms that autonomously adjust their motion in response to the environment using multistability and mechanical stimuli. By cleverly designing energy landscapes, their robot can: - Sort objects by mass - Change gait and direction when bumping into obstacles - Operate using only a single actuato

A team, including researchers from MIT AeroAstro, is designing robotic helpers that can learn from demonstration through a three-in-one training interface. Their method enables a person to lead a robot through a task through any of three training methods: via remote control; by physically moving the robot through the motions; or by performing the task themselves while the robot watches and mimics. Video by CSAIL. https://news.mit.edu/2025/new-tool-gives-anyone-ability-to-train-robot-0717 | MIT S