A parallel gripper that automatically compresses and tilts upon contact with a surface, allowing it to scoop up thin objects without the need for complex sensors or control systems.
Background and Necessity of the Invention
With the rapid expansion of e-commerce and smart manufacturing, robots are increasingly taking over the task of picking and moving items in warehouses and on production lines. The gripper, the device at the end of a robotic arm that physically grasps objects, represents the largest segment of the robot end-effector market, which is growing at a high annual rate of around 10%. In logistics, in particular, the ability to perform "flexible grasping"—quickly and safely picking up a vast array of items (SKUs) with varying shapes and packaging—has emerged as a key challenge for automation.
The most widely used type is the parallel gripper, which grasps objects by closing two jaws in parallel. However, standard jaws made of rigid materials, designed for a strong grip, are precise only for objects of a specific shape; they struggle to engage properly when the shape changes even slightly. This necessitates changing jaws for different objects, limiting their use to standardized environments. While soft or Fin-ray grippers have been developed to adapt to object shapes using flexible materials, they are limited by low repeatability and weak gripping force.
Picking up thin objects lying flat on a surface, such as business cards, paper, or sheets, is particularly challenging. Robotics researchers consider picking thin objects off a flat, hard surface with a two- or three-fingered gripper a classic "hard problem." Rigid fingertips cannot wedge themselves between the object and the surface, and applying excessive force often leads to collisions between the object, the surface, and the fingertip, resulting in a failed grasp. Existing solutions involve adding vision cameras and force/torque sensors or moving the robotic arm to tilt the object, but these methods increase system complexity, control requirements, and costs.
Furthermore, grippers using rotary finger mechanisms often have their center of rotation at the fingertip when in the initial, fully extended position, preventing flexible interaction with the surrounding environment, such as the floor. Consequently, there was a need for a new grasping structure that could automatically compress and tilt upon contact with the floor, adapting to the surface to scoop up thin objects without the need for expensive sensors or complex control logic.
Technical Principles and Implementation
This invention imparts "compliance" to the fingertips using only mechanical links and springs, without the need for additional motors, mimicking the way a human finger relaxes and adjusts its angle upon touching a surface. The gripper consists of a pair of opposing jaws (first and second jaws) mounted on a parallel drive module that closes horizontally to grasp objects. Each jaw is composed of a contact surface that touches the object/floor, a first drive mechanism that pushes the fingertip vertically, a second drive mechanism that rotates the fingertip, and a potentiometer to measure the degree of compression.
The contact surface is a flat plate positioned vertically to the ground, which makes direct contact to grasp the object. The first drive mechanism consists of an upper and lower segment connected at a specific angle, joined by a torsion spring. When the fingertip touches the floor and the arm continues to descend, the floor's reaction force causes the two segments to fold, reducing the angle and allowing the contact surface to passively retract upward. In essence, instead of pushing against the floor, the fingertip "shrinks" to adapt vertically. This structure is designed to allow the fingertip to move smoothly in one degree of freedom (up and down), protecting both the object and the device from collisions with the surface.
The second drive mechanism consists of a link unit (upper and lower links) forming a parallelogram, two pressure points (first and second) that press the fingertip, and a tension spring. The key is a two-stage operation. When both pressure points press the fingertip simultaneously, the contact surface maintains a vertical posture to grasp the object firmly. If the arm descends further so that only the second pressure point presses the fingertip, the contact surface rotates while remaining in contact with the floor, tilting at an angle. This allows the tilted fingertip to slide under the object, enabling a "scooping" motion to lift items off the floor.
The degree of fingertip pressure is indicated by the "deflection gap" between the upper end of the upper joint and the lower end of the contact part. As the hand descends, this gap narrows in stages (from wide to narrow); initially, it grips vertically, but as pressure increases, it transitions into a "wedging" posture where the fingertip tilts. A potentiometer attached to the upper joint reads this gap in real-time, allowing the system to determine the shape of the floor or object the fingertip is touching and control the chuck's position without the need for a camera. When the fingertip lifts off the surface, the restoring force of two springs returns the contact part to its original vertical position.
A parallel drive module brings a pair of these chucks together horizontally to grip an object with uniform force. Because the two chucks adapt independently, they can adjust their respective deflection gaps to conform to the shape of sloped surfaces, polygons, or irregular objects. Furthermore, the chucks and drive module are easily detached and reattached, facilitating maintenance and allowing a single chuck to be used across multiple robots.
Advantages and Expected Effects
Advantages The primary strength is its "mechanical compliance," which allows the fingertip to align with the floor without the need for separate sensors. Adaptation to the surface and the wedging motion are achieved solely through the mechanical structure of links and springs, eliminating the need for expensive force/torque sensors or complex algorithms like impedance control. As a result, the design is simple, cost-effective, and robust, while providing stronger gripping force than soft grippers and better adaptability to various shapes than rigid chucks. Since the two chucks adapt independently, they automatically conform to sloped surfaces or irregular, polygonal objects.
This represents a fundamental departure from conventional methods. Rigid grippers are precise but limited to specific shapes, while soft grippers are flexible but lack sufficient force and precision; picking up thin objects previously required adding vision systems, force sensors, and sophisticated control. This invention achieves both "strong gripping force" and "surface/shape compliance" through pure mechanical design. Additionally, the detachable structure allows for easy assembly and disassembly, aligning with the modular (plug-and-play) approach favored in modern automated facilities and enabling a single chuck to be used across various robots.
Applications A prime example is the automation of picking in e-commerce logistics. It excels at "bin picking"—retrieving items of varying shapes and sizes from warehouse bins—and is particularly effective at handling thin objects like cards, documents, flat components, and coins that are often missed or damaged by conventional grippers. It can also be widely used for tasks requiring the retrieval of objects from the floor, such as in service or sanitation robots, and for picking mixed items with polygonal or irregular shapes. Thanks to the detachable chuck structure, it can be easily integrated into various robotic arms, allowing for rapid line changes.
The ability to mechanically conform to surfaces without complex sensors or control translates to lower implementation and maintenance costs and higher reliability. It protects both the object and the device from ground impact while reliably picking up thin, irregular items that are difficult for existing automation equipment to handle. As demand grows in logistics and manufacturing for the fast, damage-free handling of objects with diverse shapes, this technology is expected to expand the utility of pick-and-place robots in unstructured environments.
Patent Listing IBL-26-0948
Inventors Professor Ki-Hun Kim, Deok-Chan Yoon, Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH)
