Achievement

Biological and robotic visual systems must be able to judge physical stability and equilibrium, which allows them to predict how objects will behave, such as whether an object is about to fall off a surface. Trainees Cholewiak, Wilder, Ringstad, and Kim are investigating the perception of the stability of 3D objects, and its dependence on shape and binocular 3D cues. They compared perceived critical angle of stability for truncated cones to physical predictions derived from geometric analysis of center of mass. Results show that human observers can make extremely precise judgments, but were aided significantly by 3D stereo cues, despite the fact that information needed for center of mass computations was available with pure 2D projections. These results indicate unexpected benefits to a global stereo representation of simple shapes that may play important roles in motor control (human or robot).