Extended Data Fig. 3: Multi-object landscape.
From: Embedding high-resolution touch across robotic hands enables adaptive human-like grasping
We examine the grasp relationships among six example objects (pawn, vase, multimeter, board eraser, coffee bottle, and Coke can) using a large disconnectivity graph. This landscape comprises 79 basins, each categorized into one of three grasp types (Power, Intermediate, Precision) based on the majority of grasps it contains. Within each basin, similar grasp strategies are observed across different objects, as illustrated in (a)-(f).
