Researchers at Shanghai Jiao Tong University in Beijing have developed a cutting-edge robotic hand that mimics human-like coordination among its palm and fingers.

This new technology allows the hand to grasp objects securely while also sensing their three-dimensional shapes and distinguishing between various materials, overcoming longstanding challenges in the field of robotics.

The innovation addresses the often overlooked role of the palm in manipulation.

Traditionally, research into soft robotic hands has concentrated on enhancing finger sensitivity, neglecting the palm, which is essential for mechanical support and sensory feedback.

The research team has successfully integrated both components for seamless interaction, achieving levels of sensitivity and control that surpass previous designs.

Central to this advancement are three key features: a high-resolution tactile palm equipped with 181,000 sensing units per square centimeter—754 times more sensitive than human skin—highly flexible fingers with a bending capacity of 204.3 degrees, and intelligent coordination algorithms.

These components work in synchronization to optimize performance.

The palm integrates a sophisticated visual-tactile system that combines micro-cameras with multi-layer elastic sensors, enhancing its ability to perceive and respond to its environment.

Meanwhile, the fiber-reinforced pneumatic fingers provide both strength and gentleness, capable of exerting a grasping force of 14.6 newtons.

In practical demonstrations, the robotic hand has exhibited capabilities comparable to human dexterity.

It has successfully performed intricate tasks, including the precise retrieval of cards, detecting minute flaws in textiles, and even pouring tea—with the hand adapting its grip to prevent any spillage.

Further augmenting its functionality, the system incorporates advanced artificial intelligence, significantly boosting its object recognition capabilities.

In a series of blind tests, the robotic hand displayed exceptional accuracy in identifying and retrieving specific items from opaque bags containing a mix of small objects, ranging from peas and sunflower seeds to larger nuts and bolts.

The system achieved an overall recognition accuracy rate of 88 percent, with a perfect success rate for identifying more distinctive objects, such as sunflower seeds.

The research team highlighted the potential of integrating rich palm tactile sensing with agile finger movement for enhanced interactive robotic functions.

Their findings have been published in the scientific journal Nature Communications.