Advanced robotic manual dexterity

 


The project

Yaskawa Italia and ARTES 4.0 launched in the summer of 2021 a partnership aimed at carrying out activities to promote, design and implement technological innovation, research and development and 4.0 projects in industry. It is a partnership strongly desired by both entities, which share an awareness of how, very often, projects of this kind are fundamental to technological progress as they make it possible to develop application solutions and address issues for which the market is not yet mature and which would otherwise struggle to find the necessary investment from companies.

The first fruit of this partnership is an interesting application prototype. "It is a telemanipulation demonstrator. With it, the operator is able to remotely move a bimanual robot to perform dexterity tasks," explains Prof. Antonio Frisoli, president of ARTES 4.0 and professor of Robotics at the Scuola Superiore Sant'Anna in Pisa, where he directs the Human-Robot Interaction area at the Percro laboratory.

On the hardware side, the system is based on a 15-axis, dual-arm Motoman SDA robot equipped with robotic hands with fine manipulative capabilities that can grasp and move objects in a human-like manner. The arms can work synchronously or perform different operations simultaneously. An operator can control a specially designed interfacing system. The solution is then completed with a vision system.

Relevant issues

The research project presented two particularly challenging issues: on the one hand, a more software aspect, related to middleware and 'systems integration, and on the other hand, an issue more directly related to the human-robot interfacing systems needed to be able to realize telemanipulation of parts.

On the first front, it was necessary to prepare an intermediate layer capable of ensuring the reliability and security of communication.

"With the contribution of the in-house embedded experts at ARTES 4.0, we developed an intermediate element that allows us on the one hand to consistently communicate with the robot through the Yaskawa API libraries and on the other hand to interface with all peripherals by managing a dedicated real-time scheduling. In this way we were able to ensure the reliability, safety and performance of the robot, but at the same time also the correct interaction for example with the vision system, which performs image processing with AI algorithms," Frisoli elaborates.

Regarding the human-robot interface, the choice was to use force-feedback interfaces that are characterized by their very small footprint and intuitiveness of use. The chosen interfaces allow the robots to move through space in three degrees of freedom and have a dedicated knob to operate the grip and manage it to fit the object to be grasped.

Almost human-like movements

"To achieve high grasping performance, we adopted two different hands, both developed by the Institute of BioRobotics of the Scuola Superiore Sant'Anna, which is one of the members of ARTES 4.0," Frisoli continues.

One arm has in fact been equipped with an under-actuated hand, capable of adapting its shape to the object to be grasped and distributing the force over the various phalanges to provide the highest quality of grip. The other, on the other hand, has a prosthetic hand, optimized to provide grip stability with low power consumption. The combination of the two offers the operator the possibility of having a hand that better fits the grip and one that instead ensures greater reliability of the grip itself over time.

With its almost human-like flexibility of movement, the Motoman SDA series was a suitable tool to carry out the ARTES 4.0 research project. "We were working on bimanual telemanipulation, and we really liked the idea of being able to have a robot that had the necessary capabilities to replicate complex bimanual tasks, which is why we turned to a demonstrator of this type," Frisoli recalls.

Open potentials

The adoption of an intelligent vision system can also provide the application with an additional degree of autonomy. In fact, the system can operate in fully operator-assisted mode, but also in autonomous or semiautonomous mode. "The possibilities are diverse: the system can identify potential objects suitable for gripping and signal them to the operator who then makes the final choice, but it can also manage the initial phase of the task and leave only the fine part of gripping to the operator, relieving him of the burden of moving the robot to the gripping point.

Obviously, in the case where there is an interaction between the two hands, the complexity increases and the operator's role is clearly more relevant: he has a coordination and assembly task," explains Frisoli, illustrating how a hybrid mode is particularly useful for handling a task that has some more standard phases and others that are more articulated and more complex.

"In reality, the challenges are very different depending on the task at hand-an assembly is extremely different from a sewing task rather than a remote medical examination," Frisoli concludes. However, the solution already represents an effective foundation that can then be customized according to the specifics of use, which can highlight the possibilities and advantages of using advanced manipulation solutions in extremely varied application landscapes.