The integration of robots into human environments is advancing rapidly, driven by the demand for systems that combine robot accuracy and repeatability with human flexibility and adaptability. In this context, human-centered manipulation applications must address uncertainties arising from human, robotic, and environmental factors. As a result, effective robotic manipulation requires both accurate pre-grasping motions and precise grip strength control, especially in tasks where robotic devices are remotely controlled to grip objects with fine, desired, and adjustable grip force. The present work tackles the challenges posed by uncertainties and non-ideal conditions in surface electromyography (sEMG)-driven human-in-the-loop (HITL) robot hand control applications. In this regard, a novel probabilistic shared autonomy framework for fine grip strength regulation is introduced, leveraging Hidden Markov Models (HMMs) applied to tactile data to encode the HITL grasping action into proper, probabilistically consistent phases. These phases are then exploited to modulate the level of shared autonomy between the human operator and the robot hand, enabling precise control over grip strength. The presented shared autonomy framework was evaluated under multiple experimental conditions, testing grip force regulation performance with a group of 10+10 intact-limb participants and a participant with amputation in both static (fixed hand-object configuration) and dynamic (pick-and-place and recipe preparation) grasping tasks, with differentiated goals inspired by real-world requirements. Moreover, to explore generalizability, experiments were conducted with both anthropomorphic and industrial robotic hands, properly equipped with tactile sensors. Experimental outcomes are supported by statistical analysis and show, for the considered sample, the effectiveness of the proposed shared autonomy control architecture in achieving fine, smooth, and controllable grip strength regulation with respect to the baseline case in absence of our approach.

Myocontrolled robotic hands require accurate and responsive control to regulate grasp strength effectively. However, many human-in-the-loop (HITL) control systems still lack robust closed-loop solutions for fine grip force regulation, limiting their performance. This paper presents a novel control system for myocontrolled hands that combines contact force sensing and vibrotactile feedback to enable more natural and precise grasp interaction. The system features an advanced force controller based on fuzzy logic, with parameter optimization guided by user preferences collected through a graphical user interface (GUI) using Global Learning of Input–Output Strategies from Pairwise Preferences (GLISp). It is compared against heuristic model and neural network based controllers. The system was validated through real-world experiments using the AR10 robotic hand with OptoForce fingertip sensors, demonstrating improved adaptability and fine force regulation capabilities for the user.

Deformable Multi-Linear Objects (DMLOs), or Branched Deformable Linear Objects (BDLOs), are flexible objects that possess a linear structure similar to DLOs but also feature branching or bifurcation points where the object’s path diverges into multiple sections. The representation of complex DMLOs, such as wiring harnesses, poses significant challenges in various applications, including robotic systems’ perception and manipulation planning. This paper proposes an approach to address the robust and efficient estimation of a topological representation for DMLOs leveraging a graph-based description of the scene obtained via graph neural networks. Starting from a binary mask of the scene, graph nodes are sampled along the objects’ estimated centerlines. Then, a data-driven pipeline is employed to learn the assignment of graph edges between nodes and to characterize the node’s type based on their local topology and orientation. Finally, by utilizing the learned information, a solver combines the predictions and generates a coherent representation of the objects in the scene. The approach is experimentally evaluated using a test set of complex real-world DMLOs. Within an offline evaluation, the proposed approach achieves a Dice score exceeding 90% in predicting graph edges. Similarly, the identification accuracy ofbranchandintersectionpoints in the graph topology is above 90%. Additionally, the method demonstrates efficient performance, achieving a runtime of over 20 FPS. In an online assessment employing a dual-arm robotic setup, the approach is successfully applied to disentangle three automotive wiring harnesses, demonstrating the effectiveness of the proposed approach in a real-world scenario.

This paper proposes a robotic based system, equipped with suitably developed mechatronic tools, able to fine manipulate medical hoses directly from the production line and during the execution of quality check tasks. The system combines suitably developed technologies (i.e., tactile sensors and cutting machine) with methodological approaches (i.e., tactile indicator, blind search methods, image processing) in order to automatize the manipulation of very thin and soft linear deformable objects, where the main challenge is the very small forces involved during manipulation. The work presents the system testing during the execution of two real quality check scenarios: verification of the hose internal diameter and checking of the hose section shape and dimensions. For the internal diameter check, a suitably defined contact indicator based on tactile signals is proposed, and it is combined with a blind search method, based on the use of a spiral path, in order to generalize the task execution, by overcoming limitations due to the possible misalignment among grasped area and hose end. For the inspection of the hose section, a specific cutting tool has been developed and image processing methods have been implemented, by means of microscope feedback, for the centering and focus of the hose sample to be analyzed.

The robotic manipulation of deformable objects, such as clothes and fabric, is known as a complex task from both the perception and planning perspectives. Indeed, the stochastic nature of the underlying environment dynamics makes it an interesting research field for statistical learning approaches and neural policies. In this work, we introduce a novel attention-based neural architecture capable of solving a smoothing task for such objects by means of a single robotic arm. To train our network, we leverage an oracle policy, executed in simulation, which uses the topological description of a mesh of points for representing the object to smooth. In a second step, we transfer the resulting behavior in the real world with imitation learning using the cloth point cloud as decision support, which is captured from a single RGBD camera placed egocentrically on the wrist of the arm. This approach allows fast training of the real-world manipulation neural policy while not requiring scene reconstruction at test time, but solely a point cloud acquired from a single RGBD camera. Our resulting policy first predicts the desired point to choose from the given point cloud and then the correct displacement to achieve a smoothed cloth. Experimentally, we first assess our results in a simulation environment by comparing them with an existing heuristic policy, as well as several baseline attention architectures. Then, we validate the performance of our approach in a real-world scenario. Project website: link

Deformable Objects (DOs), e.g. clothes, garments, cables, wires, and ropes, are pervasive in our everyday environment. Despite their importance and widespread presence, many limitations exist when deploying robotic systems to interact with DOs. One source of challenges arises from their complex perception. Deep learning algorithms can address these issues; however, extensive training data is usually required. This paper introduces a method for efficiently labeling DOs in images at the pixel level, starting from sparse annotations of key points. The method allows for the generation of a real-world dataset of DO images for segmentation purposes with minimal human effort. The approach comprises three main steps. First, a set of images is collected by a camera-equipped robotic arm. Second, a user performs sparse annotation via key points on just one image from the collected set. Third, the initial sparse annotations are converted into dense labels ready for segmentation tasks by leveraging a foundation model in zero-shot settings. Validation of the method on three different sets of DOs, comprising cloth and rope-like objects, showcases its practicality and efficiency. Consequently, the proposed method lays the groundwork for easy DO labeling and the seamless integration of deep learning perception of DOs into robotic agents.

Recently, electroencephalographic (EEG) signals have been used to enhance Human-Robot Interaction (HRI). In particular, Error-Related Potentials (ErrPs) have been exploited since very few years. These potentials are evoked when there is a mismatch between the command given by the subject and the movements of the robot, or if the user’s expectation is different from the robot or other human behavior. These signals can be used to improve and customize the robot system, as feedback to better adapt the robot to human needs. This work aims to investigate and detect the ErrPs during different interaction tasks. We set up an experiment divided into five different tasks, where every task has 120 events with a 25%-35% probability of error. The robot used in the experiment is a Baxter robot and the commands from the subject to the robot are sent in two different ways: with a keyboard or with a motion capture device. This work aims to reproduce a simplified teleoperated pick and place task. However, the achieved results do not allow to correctly identify the ErrPs, but exhibit only some minor differences between trials with and without errors. Hence, we here analyze the reasons behind such negative results, focusing on the challenges of the structure and the setup of the experiment. We analyze the possible problems and provide some recommendations to overcome them in similar use cases.

Deformable Linear Objects (DLOs) such as cables, wires, ropes, and elastic tubes are numerously present both in domestic and industrial environments. Unfortunately, robotic systems handling DLOs are rare and have limited capabilities due to the challenging nature of perceiving them. Hence, we propose a novel approach namedRT-DLOfor real-time instance segmentation of DLOs. First, the DLOs are semantically segmented from the background. Afterward, a novel method to separate the DLO instances is applied. It employs the generation of a graph representation of the scene given the semantic mask where the graph nodes are sampled from the DLOs center-lines whereas the graph edges are selected based on topological reasoning.RT-DLOis experimentally evaluated against both DLO-specific and general-purpose instance segmentation deep learning approaches, achieving overall better performances in terms of accuracy and inference time.

Trajectory learning is one of the key components of robot Programming by Demonstration approaches, which in many cases, especially in industrial practice, aim at defining complex manipulation patterns. In order to enhance these methods, which are generally based on a physical interaction between the user and the robot, guided along the desired path, an additional input channel is considered in this article. The hand stiffness, that the operator continuously modulates during the demonstration, is estimated from the forearm surface electromyography and translated into a request for a higher or lower accuracy level. Then, a constrained optimization problem is built (and solved) in the framework of smoothing B-splines to obtain a minimum curvature trajectory approximating, in this manner, the taught path within the precision imposed by the user. Experimental tests in different applicative scenarios, involving both position and orientation, prove the benefits of the proposed approach in terms of the intuitiveness of the programming procedure for the human operator and characteristics of the final motion.

In this article, the variety of approaches proposed in the literature to address the problem of mapping human to robot hand motions are summarized and discussed. We particularly attempt to organize under macrocategories the great quantity of presented methods that are often difficult to be seen from a general point of view due to different fields of application, specific use of algorithms, terminology, and declared goals of the mappings. First, a brief historical overview is reported, in order to provide a look on the emergence of the human to robot hand mapping problem as a both conceptual and analytical challenge that is still open nowadays. Thereafter, the survey mainly focuses on a classification of modern mapping methods under the following six categories: direct joint, direct Cartesian, task-oriented, dimensionality reduction based, pose recognition based, and hybrid mappings. For each of these categories, the general view that associates the related reported studies is provided, and representative references are highlighted. Finally, a concluding discussion along with the authors' point of view regarding future desirable trends are reported.

In this article, an innovative algorithm for instance segmentation of wires called Ariadne+ is presented. Although vastly present in many manufacturing environments, the perception and manipulation of wires is still an open problem for robotic applications. Wires are deformable linear objects lacking of any specific shape, color, and feature. The proposed approach uses deep learning and standard computer vision techniques aiming at their reliable and time effective instance segmentation. A deep convolutional neural network is employed to generate a binary mask showing where wires are present in the input image, then the graph theory is applied to create the wire paths from the binary mask through an iterative approach that aims to maximize the graph coverage. In addition, the B-Spline model of each instance, useful in manipulation tasks, is provided. The approach has been validated quantitatively and qualitatively using a manually labeled test dataset and by comparing it against the original Ariadne algorithm. The timings performances of the approach have been also analyzed in depth.

In this paper we present the architecture of a mobile robotized system for logistic applications. The task to be performed consists of extracting items from homogeneous pallets and, subsequently, assembling new pallets with heterogeneous goods. A further requirement is the capability of a safe human-robot interaction. For such purposes, we employ an autonomous mobile robot, equipped with a serial collaborative arm, a lifting mechanism, and a multi-sensor vision apparatus. The overall system is conceived to handle packages by dragging them aboard the mobile platform. Accordingly, we integrate a conveyor, whose vertical position can be adjusted by means of a scissor lifting mechanism. Thus, packages from different overlapping layers, which compose a generic pallet, can be introduced and stored on board.

In this paper, the problem of identifying optimal grasping poses for cloth-like deformable objects is addressed by means of a four-steps algorithm performing the processing of the data coming from a 3D camera. The first step segments the source pointcloud, while the second step implements a wrinkledness measure able to robustly detect graspable regions of a cloth. In the third step the identification of each individual wrinkle is accomplished by fitting a piecewise curve. Finally, in the fourth step, a target grasping pose for each detected wrinkle is estimated. Compared to deep learning approaches where the availability of a good quality dataset or trained model is necessary, our general algorithm can find employment in very different scenarios with minor parameters tweaking. Results showing the application of our method to the clothes bin picking task are presented.

This work addresses the task of robot depalletizing by means of a mobile manipulator, taking into account the problem of localizing the boxes to be removed from the pallet and a manipulation strategy that allows to pull the boxes without lifting them with the robot arm. The depalletizing task is of particular interest in the industrial scenario in order to increase efficiency, flexibility and economic affordability of automatic warehouses.The proposed solution makes use of a multi-sensor vision system and a force-controlled collaborative robot in order to detect the boxes on the pallet and to control the robot interaction with the boxes to be removed. The vision system comprises a fixed 3D Time-of-flight camera and an eye-in-hand 2D camera. Preliminary experimental results performed on a laboratory setup with a fixed-based robotic manipulator are reported to show the effectiveness of the perception and control system.

In this paper we investigate how to effectively deploy deep learning in practical industrial settings, such as robotic grasping applications. When a deep-learning based solution is proposed, usually lacks of any simple method to generate the training data. In the industrial field, where automation is the main goal, not bridging this gap is one of the main reasons why deep learning is not as widespread as it is in the academic world. For this reason, in this work we developed a system composed by a 3-DoF Pose Estimator based on Convolutional Neural Networks (CNNs) and an effective procedure to gather massive amounts of training images in the field with minimal human intervention. By automating the labeling stage, we also obtain very robust systems suitable for production-level usage. An open source implementation of our solution is provided, alongside with the dataset used for the experimental evaluation.

In this article, we propose an augmented reality semiautomatic labeling (ARS), a semiautomatic method which leverages on moving a 2-D camera by means of a robot, proving precise camera tracking, and an augmented reality pen (ARP) to define initial object bounding box, to create large labeled data sets with minimal human intervention. By removing the burden of generating annotated data from humans, we make the deep learning technique applied to computer vision, which typically requires very large data sets, truly automated and reliable. With the ARS pipeline, we created two novel data sets effortlessly, one on electromechanical components (industrial scenario) and other on fruits (daily-living scenario) and trained two state-of-the-art object detectors robustly, based on convolutional neural networks, such as you only look once (YOLO) and single shot detector (SSD). With respect to conventional manual annotation of 1000 frames that takes us slightly more than 10 h, the proposed approach based on ARS allows to annotate 9 sequences of about 35 000 frames in less than 1 h, with a gain factor of about 450. Moreover, both the precision and recall of object detection is increased by about 15% with respect to manual labeling. All our software is available as a robot operating system (ROS) package in a public repository alongside with the novel annotated data sets. Note to Practitioners-This article was motivated by the lack of a simple and effective solution for the generation of data sets usable to train a data-driven model, such as a modern deep neural network, so as to make them accessible in an industrial environment. Specifically, a deep learning robot guidance vision system would require such a large amount of manually labeled images that it would be too expensive and impractical for a real use case, where system reconfigurability is a fundamental requirement. With our system, on the other hand, especially in the field of industrial robotics, the cost of image labeling can be reduced, for the first time, to nearly zero, thus paving the way for self-reconfiguring systems with very high performance (as demonstrated by our experimental results). One of the limitations of this approach is the need to use a manual method for the detection of objects of interest in the preliminary stages of the pipeline (ARP or graphical interface). A feasible extension, related to the field of collaborative robotics, could be used to exploit the robot itself, manually moved by the user, even for this preliminary stage, so as to eliminate any source of inaccuracy.

In this paper, a tactile-based robotic system to perform the insertion of a Deformable Linear Object (DLO) in a hole is proposed. This is a typical application in manufacturing processes involving assembly of electric cables in connectors or electromechanical components. A Recurrent Neural Network (RNN) with Long Short Term Memory (LSTM) cells is adopted in this work to predict the external forces acting on the DLO from the tactile data. The tactile sensor, mounted on the finger, provides 16 signals and it is the only sensor required during the effective insertion task. In a real environment, the tight spaces very often prevent the possibility to use the vision system, also when the same task is performed by a human being. Force/Torque sensors instead increase the system price and provide signals that might be affected by inertia disturbance or other undesired effect, that are difficult to manage. The control law design is based on the RNN outputs and the distance between end-effector and target hole. In particular, it leads the plastically deformed DLO inside the hole while it adjusts the tool pose guided by the control errors in order to prevent buckling. The contribute of this work is dual: first a tactile feedback integrating a RNN to estimate contact forces on the grasped object is presented; second, a control system is developed to perform a challenging insertion of a DLO in a hole. Experimental works are presented to validate the proposed algorithms.

This paper reports the development of a manipulation system for electric wires, implemented by means of a commercial gripper installed on an industrial manipulator and equipped with cameras and suitably designed tactile sensors. The purpose of this system is the execution of wire insertion on commercial electromechanical components. The synergy between computer vision and tactile sensing is necessary because, in a real environment, the tight spaces very often prevent the possibility to use the vision system, also when the same task is performed by a human being. A novel technique to speed up the generation of training data sets for convolutional neural networks (CNNs) is proposed. Therefore, this technique is used to train a CNN in order to detect small objects (such as wire terminals). Moreover, aiming to prevent faults during the task and to interact with the environment safely, several machine learning approaches are used to produce an affordable output from the tactile sensor. The proposed approach shows how a cheap sensor embedded with suitable intelligence can provide information comparable to a more expensive force sensor.

In this paper, an ongoing work for verifying the behavior of a twisted string actuator in contact with a sliding surface or guided through a sheath is presented. The twisted string actuation system is particularly suitable for very compact and light-weight robotic devices, like artificial limbs and exoskeletons, since it allows the implementation of powerful tendon-based driving systems, based on small-size DC motors characterized by high speed, low torque and very limited inertia. One of the major limitations of this actuation system is by now related to the fact that the string should not be in contact with any obstacle, because this contact will alter the twisting angle propagation along the string and, eventually, completely stop the string twisting. This design constraint imposes a straight path between the motor and the linear load attached to the other string end. After the presentation of the basic properties of the twisted string actuation system, the model of the twisted string in contact with a sliding surface is discussed. The behavior of the system has been then experimentally verified and discussed. A preliminary evaluation of control strategies for compensating the side effects generated by the contact of the twisted string with the sliding surface is also presented.

In this paper, an ongoing work for the implementation of a variable stiffness joint actuated by a couple of twisted string actuators in antagonistic configuration is reported. The twisted string actuation system is particularly suitable for very compact and light-weight robotic devices, like artificial limbs, exoskeletons and robotic hands, since it renders a very low apparent inertia at the load side, allowing the implementation of powerful tendon-based driving systems, using as actuators small-size DC motors characterized by high speed, low torque and very limited inertia. The basic properties of the twisted string actuation system are firstly presented, and the way how they are exploited for the implementation of a variable stiffness joint is discussed. A simple control algorithm for controlling the joint stiffness and position simultaneously is discussed, and a the feedback linearization of the device is taken into account and validated in simulation.

In this work, we deal with the problem of planning a manipulation task for a robotic system composed of at least one dexterous arm and a dexterous multi-fingered hand. The goal of the local planner is to include both, the arm and the hand, in the execution of the task in a coordinated way. This is achieved by using the workspace of the hand which is computed offline. During the online planning, the current in-hand manipulation capability is evaluated taking advantage of the dimensions of the hand workspace and considering the task itself. Dynamic weights enable the computation of the instantaneous contributions of the two subsystems on the motion of the manipulated object. The method is evaluated in simulation as well as in several experiments on the real robot.

In this paper, the implementation of a variable stiffness joint actuated by a couple of twisted string actuators in antagonistic configuration is presented. The twisted string actuation system is particularly suitable for very compact and light-weight robotic devices, like artificial limbs and exoskeletons, since it renders a very low apparent inertia at the load side, allowing the implementation of powerful tendon-based driving systems, using small-size DC motors characterized by high speed, low torque and very limited inertia. After the presentation of the basic properties of the twisted string actuation system, the way how they are used for the implementation of a variable stiffness joint is discussed. A simple PID-based motor-side algorithm for controlling simultaneously both the joint stiffness and position is discussed, then the identification of the system parameters is performed on an experimental setup for verifying the proposed model and control approach.

In this paper, the design and experimental evaluation of a cable driven robotic gripper for underwater applications is presented. The gripper has three fingers and is characterised by a large workspace if compared with other similar devices reported in literature. Its kinematic configuration allows to execute both parallel and precision grasps on objects with very different dimensions. The gripper has 8 degrees of freedom actuated by only three motors by means of a suitable coupling of the joints obtained through the cable transmission. Moreover, in order to facilitate the execution of complex tasks, special force/torque sensors are mounted on the fingertips. The paper reports the main specifications deriving from the particular tasks in which the gripper is involved, and illustrates the proposed design solutions. Results obtained from real underwater experiments are provided as well, in order to demonstrate the capabilities of the gripper.

In this paper, a novel force/torque sensor is presented. The sensor is based on optoelectronic components and therefore its design is relatively simple and reliable. The sensor design make it suitable for the integration in different robotic systems, such as e.g. the fingers of robotic hands. The basic principle and the design of the sensor are described in this paper, along with a specific prototype implemented for underwater applications. Experimental data are presented and discussed to illustrate the main features of the proposed sensor, and its use as an intrinsic tactile sensor is evaluated.

This paper presents the design of an optical position sensor integrated into a miniaturized tendon-driven robotic joint. The sensor exploits the modulation of the light power flux that goes from an infrared Light Emitting Diode (LED) to a PhotoDiode (PD) by means of a variable-thickness canal integrated into the joint itself to detect the joint position. The LED and the PD are fixed on one of the two links that compose the robotic joint, while the canal is integrated into the other link. The paper reports the basic sensor working principle, the integrated design of the miniaturized robotic joint with embedded position sensor and the experimental evaluation of the proposed device on a force/position control loop. Finally, a preliminary prototype of the UBH-IV finger in which the proposed joint with embedded position sensor are used is presented.

In this paper, a method to derive the three predominant synergies and their temporal weights for planning grasps of the UB Hand IV (University of Bologna Hand, version IV) is proposed. The method adopted to define the postural synergies from experiments is based on the kinematic structure of the robotic hand and on the taxonomy of the grasps of common objects. The control strategy, exploiting postural synergies, that drives the hand during reach to grasp is further described. During prehension the hand moves continuously in a configuration space of highly reduced dimensionality with respect to its degrees of freedom. The experiments confirm that the UB Hand IV works efficiently in a synergy based framework for grasp planning and prehension control. It is shown that the introduction of the third predominant synergy significantly improves the grasping synthesis and performance, especially for the adduction/abduction motion of the thumb.

In this paper, the control of a redundant robotic manipulator with variable stiffness actuation is addressed. The problem of controlling simultaneously the end-effector position and stiffness exploiting the robot redundancy for the optimization of the robot configuration is considered, and the relation between the manipulator redundancy and the selection of both the joint and end-effector stiffness is discussed. The controller is configured as a cascade system that allows the decoupling of the actuators dynamics from the arm dynamics and the consequent reduction of the order of the manipulator dynamic model. Only the actuator and joint positions are needed by the controller, introducing in this way a significant simplification with respect to previously proposed state feedback techniques. The effectiveness of the proposed approach is verified by simulations of a 3-DOF planar manipulator.

This video presents the evolution of the robotic hands, called UB Hands (University of Bologna Hands), developed at the Laboratory of Automation and Robotics of the University of Bologna during more than 25 years of research in this field. Starting from the UB Hand I, the first robot hand prototype developed in our labs, the different design solutions and philosophies that have been followed toward the innovative UB Hand IV, also called DEXMART Hand, recently developed within the DEXMART project are presented. Remarkable characteristics of the UB Hands are also the whole hand manipulation capabilities, the ability of reconstructing the contact forces over the whole hand surface as in the case of the UB Hand II, and the presence of force/tactile sensors as in the UB Hand IV. Moreover, the use of soft covers for the emulation of the human tissue characteristics has been studied, and the adoption of innovative design concepts based on compliant structures has been introduced in the UB Hand III and IV.

In this paper, we deal with several aspects related to the control of tendon-based actuation systems for robotic devices. In particular, the problems that are considered in this paper are related to the modeling, identification, and control of tendons sliding on curved pathways, subject to friction and viscoelastic effects. Tendons made in polymeric materials are considered, and therefore, hysteresis in the transmission system characteristic must be taken into account as an additional nonlinear effect because of the plasticity and creep phenomena typical of these materials. With the aim of reproducing these behaviors, a viscoelastic model is used to model the tendon compliance. Particular attention has been given to the friction effects arising from the interaction between the tendon pathway and the tendon itself. This phenomenon has been characterized by means of a LuGre-like dynamic friction model to consider the effects that cannot be reproduced by employing a static friction model. A specific setup able to measure the tendon's tension in different points along its path has been designed in order to verify the tension distribution and identify the proper parameters. Finally, a simple control strategy for the compensation of these nonlinear effects and the control of the force that is applied by the tendon to the load is proposed and experimentally verified.

This paper presents the latest results in the development of the low-level controller of the robotic hand UBH-IV (University of Bologna Hand, version IV). In particular, the friction effects acting at joint level have been rendered by means of a LuGre-like model and a procedure for the identification of the friction model parameters is described. With the aim of providing an online estimation of the effects due to the interaction of the robotic hand with the environment, a controller able to evaluate the overall external torque acting on the finger joints and to discern between friction and torques generated by the external interaction force without using direct measures of the contact forces is proposed. The identification and control tests are carried over on an experimental setup composed by a single finger phalanx, manufactured with the same material and techniques of the hand itself.

In this paper, an innovative sensor based on optoelectronic components and compliant frames for the measurement of the tendon tension is presented. With respect to conventional solutions for force sensing, like strain-gauge or Bragg-grating based force sensors, this sensor presents several advantages, mainly in terms of compactness, simplicity of the implementation and conditioning electronics. The proposed sensor exploits the properties of optoelectronic components with a narrow angle of view to measure the very small deformation of a compliant frame caused by the tendon tension. The sensor can be placed at the tendon ends as such as in any position along the tendon. The paper reports the basic working principle and a simplified procedure for the design of the sensor frame together with the results of an experimental testbench where a couple of the proposed sensors are use for the feedback control of a tendon-driven robotic joint.

In this paper, the postural synergies configuration subspace given by the fundamental eigengrasps of the UB Hand IV (University of Bologna Hand, version IV) is derived through experiments. This study is based on the kinematic structure of the robotic hand and on the taxonomy of the grasps of common objects. Experimental results show that it is possible to obtain grasp synthesis for a large set of objects both in the case of precision or power grasps by using only a very limited set of dominant eigengrasps. The tasks here presented are planned with an initial hold of the hand followed by reach and grasp phases, that are unique for each object/grasp combination, during which the robotic hand posture evolves continuously within a subset of the hand configuration space given by the two predominant eigenpostures. The paper reports the method adopted to define from experiments the postural synergies for the UB Hand IV and the results of the grasp tasks performed adopting the defined synergies.

Viscoelastic contact interfaces can be found in various robotic components that are covered with a compliant surface (pad) such as anthropomorphic hands, biomimetic haptic/tactile sensors, prostheses, and orthoses. In all these cases, it is desirable to obtain thin and resistant pads with predetermined compliance and damping properties (e.g., mimicking the human skin and pulpy tissues). In order to overcome the limits of homogeneous layers of a soft viscoelastic material, which is commonly used in the aforementioned devices, this paper suggests the adoption of soft pads that are composed of a continuous external layer (skin) coupled with an internal layer having fluid-filled voids. The process to design the pad starts with the selection of a hyperelastic medium with proper tribological features, whose constitutive parameters are determined by numerically fitting nonlinear stress-strain curves under pure homogenous deformations. The optimization of the internal layer morphology is then achieved through nonlinear finite element analysis (FEA) that provides an estimate of hardness and friction influence on the pad static compliance. Finally, the pad is filled with a viscous fluid that is chosen to modify time-dependent phenomena and to increase damping effects. The effectiveness of the procedure is proven by designing and modeling better-behaved artificial pads that mimic human-finger dynamic properties.

This paper reports the modeling activities related to the development of an innovative tendon-driven robotic finger, designed as the fundamental element of a new biologically-inspired artificial hand. The finger is realized in plastic material by means of 3D-printing, a production process that allows a remarkable simplification of the mechanical design. Through 3D-printing, we were able to easily implement solutions that could be very difficult, if not impossible, to obtain with conventional manufacturing. A detailed simulation model of the robotic finger has been developed with the aim not only of designing and testing suitable control strategies for the finger, but also of investigating the benefits and the flaws of particular design solutions. As a matter of fact, this approach to design and realization of robotic fingers, that fulfills the requirements in terms of compactness, integration and simplified assembly, has a significant drawback in frictional phenomena on both tendons and joints. For this reason, an adapted LuGre friction model is proposed in order to simulate and study the finger behavior.

In this paper, the characterization of the force distribution along a tendon sliding on a curved pathway, subject to friction and visco-elastic effects, is investigated. In order to have a better understanding of the system behavior, a specific setup able to measure tension forces in different points along the tendon's path has been built. Experimental data collected by measuring the tendon tension forces during both the pulling and the release phase are presented, and theoretical models reproducing the tendon behavior with increasing fidelity are proposed. In particular, the friction arising from the interaction between the tendon pathway and the tendon itself is characterized by means of a LuGre-like dynamic friction model. The introduction of a dynamic friction model allows to reproduce in simulation some effects arising during experimental activities that cannot be reproduced employing an equivalent static friction model. Moreover, the adoption of tendons made by polymeric fibers introduces hysteresis in the tendon transmission characteristic due to the plasticity and creep phenomena typical of these materials. With the aim of reproducing this behavior, a visco-elastic model is used for modeling the tendon compliance.

The current research effort in the design of lightweight and safe robots is resulting in increased interest for the development of variable stiffness actuators. Antagonistic pneumatic muscle actuators (pMAs) have been proposed for this purpose, due to their inherent nonlinear spring behavior resulting from both air compressibility and their nonlinear force-length relation. This paper addresses the simultaneous torque and stiffness control of an antagonistically actuated joint with pneumatic muscles driven by compact, fast-switching solenoid valves. This strategy allows compensation of unmodeled joint dynamics while adjusting the joint stiffness depending on the task requirements. The proposed controller is based on a sliding mode force control applied to an average model of the valve-pneumatic muscle system. This was necessary to cope with both the well known model uncertainties of the pMA and the discontinuous on-off behavior of the solenoid valves. Preliminary experimental results verified the effectiveness of the proposed implementation.

Tendon-based transmission systems present many positive aspects and greatly simplify the mechanical design of small robotic devices, such as robotic fingers. On the other hand, they introduce several nonlinear effects that must be properly considered by the control algorithms to achieve a suitable performance level in the regulation of the finger joint torques. In this paper, the model of the tendons-based driving system and of the nonlinear effects arising from the use of sliding paths instead of pulleys for the tendon routing are discussed, and control algorithms aiming at compensating these nonlinearities are presented. Both models and control algorithms have been validated by experiments. In particular, in order to gain a better insight on the force distribution along the tendon, an experimental setup for the measurement of the tension in some intermediate points has been developed. After the identification of the tendon characteristics, a suitable control law for the compensation of the nonlinear effects due to the friction acting on the transmission system has been applied. The proposed compensation scheme is based on a sliding-mode controller with boundary layer, where the boundary threshold is modulated as a function of the desired tendon tension.

Physical human-robot interaction requires the development of safe and dependable robots. This involves the mechanical design of lightweight and compliant manipulators and the definition of motion control laws that allow to combine compliant behavior in reaction to possible collisions, while preserving accuracy and performance of rigid robots in free space. In this framework, great attention has been given to robots manipulators with relevant elasticity at the joints/transmissions. While the modeling and control of robots with elastic joints of finite but constant stiffness is a well- established topic, few results are available for the case of robot structures with variable joint stiffness -mostly limited to the 1-dof case. We present here a basic control study for a general class of multi-dof manipulators with variable joint stiffness, taking into account different possible modalities for changing the joint stiffness on the fly by an additional set of commands. It is shown that nonlinear control laws, based either on static or dynamic state feedback, are able to exactly linearize the closed- loop equations and allow to simultaneously impose a desired behavior to the robot motion and to the joint stiffness in an decoupled way. Illustrative simulations results are presented.

In this paper, the dynamic model of a robot with antagonistic actuated joints is presented, and the problem of full linearization via static state feedback is analyzed. The use of transmission elements with nonlinear relation between the displacement and the actuated force allows to control both the position and the stiffness of each joint. The main advantage of this actuation modality is that the achieved stiffness becomes a mechanical characteristic of the system and it is not the result of an immediate control action as in the classical impedance control scheme (Davison, 2003). Different examples of implementation of this kind of devices are known in literature, even if limited to one single joint (Kjita et al., 2003; Laumond and Kineocam, 2006; Mansard and Chaumette, 2004 and 2006) and the application of antagonistic actuated kinematic chains in the field of robotic hand design is under investigation (Stasse et al., 2006). After a brief review of the dependence of the properties of antagonistic actuation on the transmission elements characteristics, a scheme for simultaneous stiffness-position control of the linearized system is presented. Finally, simulation results of a two-link antagonistic actuated arm are reported and discussed.

In this paper, the tendon-sheath driving system for a robotic hand is presented and its force transmission characteristics are analyzed. The use of tendon-based transmission permits to reduce the size and the complexity of the actuation chain in many mechanical devices. A simple static model that describes the tendon-sheath driving system is presented, and its behavior is compared both with simulative results, obtained with a lumped parameters model of the tendon, and with experimental results. Different static and dynamic friction models are used in the simulations and the related results are compared to highlight some phenomena that are not visible from the static model of the tendon. A simple force control algorithm with feedforward friction compensation based on the static friction model is also presented

The first part of this paper describes the development of a humanoid robot hand based on an endoskeleton made of rigid links connected with elastic hinges, actuated by sheath routed tendons and covered by continuous compliant pulps. The project is called UB Hand 3 (University of Bologna Hand, 3rd version) and aims to reduce the mechanical complexity of robotic end effectors yet maintaining full anthropomorphic aspect and a good level of dexterity. In the second part this paper focuses on the early experiences of the UB Hand 3 in performing manipulation tasks.