DexWrist: A Robotic Wrist for Constrained and Dynamic Manipulation

Martin Peticco^1*, Gabriella Ulloa^1*, John Marangola¹, Pulkit Agrawal¹

MIT¹

(Under Review)
^*Indicates Equal Contribution

Motivation

The spatial orientation of an end-effector such as a hand or a gripper is closely tied to its ability to perform a desired task, and nearly all robotic and prosthetic arms incorporate some type of wrist for this purpose. Yet both the academic and industrial research communities have tended to place more focus on hand/gripper development than that of wrist systems. Recent prosthetics investigations, however, have shown that increased dexterity in wrist prostheses may contribute more to manipulation capacity than a highly dexterous terminal device with limited wrist capability. [1]

[1X Speed] DexWrist performs highly dynamic tasks with human-level wrist dexterity.

Enables more efficient and intuitive teleoperation in constrained, human-centric environments

More data efficient: Fewer environment resets required and less overall human teleoperator time in constrained spaces.

More intuitive: DexWrist's human-like kinematics and design collect shorter trajectories and more natural behavior for manipulation in constrained, human-centric environments.

Reduced horizon length in constrained environments leads to improved success rates for behavioral cloning policies

Human-like design of the wrist generates natural behavior during teleoperation resulting in faster task completion

Diffusion policies trained for the DexWrist were 3.24x faster at retrieving objects from a cluttered refrigerator than the AgileX + default wrist.

Why DexWrist?

Enables constrained, dynamic manipulation tasks on any robot arm
Torque transparency and easily simulatable kinematics enables dynamic policy learning
Facilitates faster teleoperation for more scalable data collection
Reduces trajectory lengths, making policy learning more efficient

A novel 2-DOF Parallel Kinematic Mechanism (PKM) achieves human-like kinematics while maintaining a one-to-one mapping between the two actuators and each of their DOF, making teleop and policy learning easier. The final DOF is achieved by mounting the wrist to a motor.

Custom Quasi-Direct Drive (QDD) actuators with brushless motors and a 13:1 planetary gearbox allows for dynamic tasks due to its backdriveability and speed, much how humanoid robots achieved dynamic movement using such actuators.

[1x Speed] Finally, DexWrist has a large workspace and fits on most commercial robotic arms (AgileX shown). This makes fast, dynamic tasks possible on any arm.

Join Waitlist to Purchase a DexWrist!