A Little Wiggle Helps Robots Learn the Ropes
Carnegie Mellon University researchers have developed "Wiggle and Go," a system that enables robots to estimate the physical properties of ropes and cables by observing how they respond to initial perturbations, allowing the robot to plan manipulation tasks without trial-and-error learning. The method uses computer vision to track wave propagation through the rope and estimates properties like stiffness and weight to predict behavior in applications ranging from manufacturing to search and rescue operations. The team trained their model entirely in simulation and validated it on physical robots, addressing the challenge of sim-to-real transfer for deformable object manipulation.














