5-finger hands are expected to open up a wide range of possibilities, particularly in assistance and service robotics.

© Schunk

While grippers for industrial automation have so far primarily been designed for robustness, durability and performance, gripper hands focus on the aspect of flexibility of movement. The closer humans and robots work together, the greater the relevance of humanoid 5-finger hands. "In the extreme, humans and service robots will share one and the same workplace, including all tools and aids," says Dr. Martin May, Head of Research/Advanced Technologies at Schunk, with conviction. This is precisely why the Lauffen-based company had its SVH-5 finger hand certified by the DGUV for collaborative operation back in 2017. With the help of a total of nine drives, the five fingers can perform a wide variety of gripping operations. In addition, numerous gestures can be displayed, which facilitates visual communication between humans and service robots and can also increase acceptance for use in a human environment.

Desk sharing with the robot

"In our research projects, we found that the human hand is much more than just a highly flexible instrument for manipulation. In contrast to industrial grippers, users always associate emotional aspects with humanoid gripping hands," explains May, adding: "Gripping hands are always in demand when a robot needs to imitate human behavior." This applies to manipulation as well as gestures. Current research projects are focusing in particular on household-related applications in service robotics and assembly-related applications in industrial assistance robotics.

Different gripping hands are now available to suit the respective application, ranging from a 2-finger hand for service robotics that is reduced to the basic functions of gripping to industrial 3-finger hands and complex 5-finger hands. Schunk's latest model, the so-called SIH, also has five human-like fingers, but differs fundamentally from the aforementioned SVH in terms of drive and kinematics. While the latter is driven by nine motors and fulfills the typical aspects of a precisely working robotic hand, the SIH, which is equipped with five motors and operated by cable pulls, is much more closely based on its human model with its tendons and muscles. Three of its fingers can be moved independently of each other, while the two smallest ones work together as a team.

According to Schunk, this makes the SIH more flexible to use than other gripping hands with cable pull mechanisms on the market and also more robust. In order to achieve the goal of an affordable, flexible and easy-to-use 5-finger hand, the developers have drawn on their experience in bionics as well as modern motor and electronics concepts. With the help of intelligent gripper control, a wide range of gripping processes can be implemented via an easy-to-use interface without having to program them precisely.

In this application, the SVH-5 finger hand autonomously grips any object that has been placed anywhere.

© Schunk

In its Smart Labs, Schunk is going even further: in addition to the gripping component, it is also focusing on the gripping process as a whole and looking for ways to complete handling tasks autonomously. The complex programming of the robot, which has so far been carried out manually by the user or integrator, is to be replaced in future by a learning, autonomous component network. Instead of defining positions, speeds and gripping forces step by step, intelligent gripping systems will in future detect their target objects via cameras and take over the gripping planning independently. Based on data sets and algorithms, gripping systems will be able to recognize regularities and derive appropriate reactions. Researchers are also working on algorithms to classify different geometries and arrangements and develop optimal gripping strategies. In this way, gripping systems should ultimately be able to handle parts independently and continue to refine the underlying gripping processes.

Evaluate gripping quality independently

Dr. Martin May, Schunk: "With the help of artificial intelligence methods, flexible gripping hands can continuously adapt to new objects and contexts.

© Schunk

The greater the variance of the parts to be gripped and the more complex the task, the more likely it is that gripper hands will be used. Using appropriate sensors in the gripper fingers, the motor current and intelligence integrated into the gripper hand, it should be possible to record and evaluate the quality of a grip and readjust it if necessary. In addition, object characteristics such as geometry, size or flexibility can be recorded via the gripper alone and transmitted to higher-level systems or upstream or downstream stations. "With the help of artificial intelligence methods, it will also be possible to train service and assistance robots intuitively and to create and enrich individual libraries for gripping planning," Martin May is convinced. "Flexible gripping hands in particular will no longer be used only for repetitive tasks, but will be able to continuously adapt to new objects and contexts and constantly optimize their gripping strategies."