In robotics, a gripper (end effector) is the device fixed to the end of the robot arm that is supposed to interact with the work environment and grasp the intended objects. The most commonly used grippers come in the following four types:
- Astrictive: suction forces applied to the object’s surface using vacuum, magneto, electro
- Contigutive: direct contact with objects for adhesion to take place using surface tension
- Impactive: based on jaws or claws that physically grasp by direct impact upon the object
- Ingressive: based on pins or hackles that physically penetrate the surface of the object
The Evolving Gripper Demands from E-Commerce
Although these different types of grippers are gradually evolving and their features are improving to handle more varieties of objects, they still have not reached the point of being highly effective and accurate, especially in the case of item picking in warehouses, where they are heavily used by e-commerce companies. As the e-commerce business develops further, it has become necessary for major companies to install robots in their warehouses that can replace human workers and perform faster and more accurate item picking and packing jobs. In 2015, Amazon.com started the “Amazon Picking Challenge,” inviting the academic robotics community to design robots that can grab items from warehouse shelves and place them in bins, just as human workers are doing in their warehouses. This annual challenge calls on roboticists from all over the world to compete in developing picking and stowing task solutions using advanced robotics. Humans can pick around 400 random items per hour at full speed, while the winners of the challenge can clock approximately 100 items per hour with a failure rate nearing 16%.
Usually, robotic item picking occurs in three main types of environment: structured, semi-structured, and random, with each presenting an increasing level of application complexity, cost, and cycle time. In a structured environment, parts are positioned or stacked in an organized, predictable pattern, so they can be easily imaged and picked. In a semi-structured environment, parts are positioned with some organization and predictability to help aid imaging and picking. In the random environment, parts are in totally random positions, including different orientations, overlapping, and even entangled, further complicating the imaging and picking functions. Structured and semi-structured picking are often not so complicated to achieve and can be implemented with most of the technologies in the marketplace, but when it comes to robotic picking in the random environment, then much more accuracy is needed in the face of chaos. Robotic gripping is still a widespread challenge for varieties of randomly arranged items. Some advancements are occurring in developing collaborative, adaptive, and soft grippers capable of picking random e-commerce items.
New Grippers Addressing Gaps in Robotic Item Picking
RightHand Robotics, a startup based in Somerville, Massachusetts, is developing its new RightPick gripper that combines astrictive (vacuum) and impactive (claws) gripper types in one system. According to the company, the time to value for RightPick can be demonstrated in a matter of hours, as RightHand offers rapid setup, remote support, and easy integration. RightPick is able to quickly demonstrate value in a wide variety of workflows, such as sorting batch-picked items, picking items from automatic storage and retrieval systems (AS/RS), inducting items to a unit sorter, order quality assurance, and more.
“The supply chain of the future is more about pieces than pallets,” according to RightHand Robotics Co-Founder Leif Jentoft. “RightHand can help material handling, [third-party logistics] (3PLs), and e-commerce warehouses lower costs by increasing automation.”
German gripping systems specialist SCHUNK offers its Co-act (which stands for collaborative actuator) gripping technology for future human/robot collaboration. The modular JL1 Co-act gripper uses built-in sensors to track the proximity of humans to create a safety “bubble.” By triggering evasive action, it avoids human-machine contact. The module can grip, handle, and assemble objects of any shape. Its jaws can measure gripping forces and it has tactile sensors, making its operation adaptive and responsive. The built-in sensors include two cameras that allow the gripper to see its surroundings in three-dimension (3D) and to detect items. In the future, SCHUNK Co-act grippers will be able to transmit all relevant data about processes and surroundings.
According to Dr. Wolfgang Wahlster, CEO of the German Research Center for Artificial Intelligence (DFKI), “SCHUNK’s JL1 is a perfect example of a state-of-the-art, smart-end effector.”
The Smart Grasping System from Shadow Robot Company has built-in sensors that monitor the torque in the springs of the joints and an intelligent system that knows what it is grasping as it approaches the object, therefore, it can select the correct grasping method for the object.
According to Managing Director Rich Walker, “When companies have deployed robots in the past, they’ve used grippers that were designed to perform one function, so they get a robot that performs one function. We’ve had 20 years of experience building human-like robotic hands, so it was natural for us to go along the lines of how can we make this pick up lots of different things, as a human would be able to.”