6N-DoF Pose Tracking for Tensegrity Robots

Tensegrity robots, which are composed of compressive elements (rods) and flexible tensile (e.g., cables), have a variety advantages, including flexibility, low weight, resistance to mechanical impact. Nevertheless, the hybrid soft-rigid nature these robots also complicates ability localize track their state. This work aims address what has been recognized as grand challenge in this domain, i.e., state estimation tensegrity through marker-less, vision-based method, well novel, on-board sensors that can measure length robot’s cables. In particular, an iterative optimization process is proposed 6-DoF pose each rigid element robot from RGB-D video endcap distance measurements cable sensors. To ensure estimates physically feasible, they not resulting collisions between rods or with environment, physical constraints introduced during optimization. Real-world experiments performed 3-bar robot, performs locomotion gaits. Given ground truth data motion capture system, method achieves less than 1 cm translation error 3 $$^\circ $$ rotation error, significantly outperforms alternatives. At same time, approach provide accurate throughout motion, while often fails due occlusions.