SO-ARM 102 is Robonine’s open-source leader–follower robot arm kit for teleoperation and imitation learning. It builds on the SO-ARM 100 and 101 (see the SO-101 entry) with a stiffer frame printed in carbon-fiber-filled PET and a parallel gripper in place of the angular jaw. This entry links directly to Robonine’s own repository.
The mechanism
A serial chain of five revolute arm joints ends in a parallel gripper whose two jaws are prismatic joints. The second jaw is defined to follow the first through a mimic joint, so the gripper is one actuated degree of freedom and the arm has six in all. There are no closed kinematic loops. It is a desktop arm, so it is fixed-base.
About this model
Robonine generated the URDF from its CAD design and published the checks it ran on the export alongside it: that every inertia tensor is physical, that each joint’s zero lies within its limits, and that the meshes match the CAD bounds. The arm’s collision shapes are simplified boxes, and its masses are computed from the CAD volumes, the print settings and the servo weights.
About this file — read before importing
This links directly to the simulation/urdf folder of Robonine’s
SO-ARM-102 repository. ARMOR links directly to that
repository rather than re-hosting anything, so what you get is exactly what Robonine
publishes, unmodified.