Industrial robots often rely on specialized end-effectors such as suction cups or parallel-jaw grippers, but no single gripper can reliably satisfy every constraint across a complex manipulation sequence. A new arXiv preprint by Sukhvansh Jain and colleagues asks whether two arms equipped with different simple grippers can collectively handle tasks that neither gripper could perform alone.

The authors formulate embodiment-aware execution as a finite-horizon optimization over arm-trajectory assignments. Given a single human demonstration of the desired object motion, the planner decides at each segment whether to use one arm or a coordinated two-arm handover. It jointly reasons about grasp feasibility, motion constraints, manipulability, and cross-gripper compatibility, with explicit switching regularization to produce globally consistent assignments.

Experiments on a dual Franka FR3 platform with suction and parallel-jaw grippers showed improved feasibility and reliability on tasks that were infeasible under single-gripper execution. The results suggest that heterogeneous bimanual setups can expand robotic capability without requiring more complex end-effectors.