DEX-EE hand helps Google DeepMind advance robotic manipulation

Shadow Robot’s DEX-EE hand redefines robotic manipulation by breaking conventions in design and functionality. Unlike traditional robotic hands, DEX-EE was built to be reliable and robust from the ground up.

The hand has non-human kinematics that the company said provides greater range of movement while ensuring durability. Unlike fragile robotic hands, DEX-EE’s tendons can withstand harsh impacts, even if multiple tendons are stressed simultaneously. Its modular design allows for quick replacement of fingers or parts, ensuring minimal downtime.

The hand’s stereo camera array in each fingertip enables high-resolution force sensing, from as little as 0.01N up to 18N. Two EtherCAT nodes at the base of each finger capture fine-grained tactile data, further enhanced by magnet-based Hall Effect sensors on the middle and proximal phalanges. These provide precise feedback during grasping and manipulation, making the hand exceptionally responsive.

DEX-EE also features an unconventional five-motor, four-joint design that eliminates backlash and prevents tendon slippage to provide a consistent performance. The deliberate three-finger setup allows for realistic object handling, enabling the hand to collide naturally with objects before adjusting its grip.

Shadow Robot originally developed the hand for Google DeepMind. With 12 degrees of freedom (DOF) at two-thirds the cost of its predecessor, DEX-EE is poised to bring advanced robotic manipulation to a wider market while pushing the limits of what’s possible in robotic hands.

Explore the RBR50 Robotics Innovation Awards 2025