Visual sensing of continuum robot shape using self-organizing maps

摘要: Shape control of continuum robots requires a means sensing the curved shape robot. Since are deformable, they take on shapes that general curves in space, which not fully defined by actuator positions. Vision-based shape-estimation provides promising avenue for shape-sensing. While this is often facilitated fiducial markers, sometimes fiducials feasible due to either robot's application or its size. To address this, we present robust and efficient stereo-vision-based, shape-sensing algorithm does rely assume orthogonal camera placement. The employs self-organizing maps triangulate three-dimensional backbone curves. Experiments with an object known demonstrate average accuracy 1.53 mm 239 arc length curve.