We present admittance control and fingertip contact detection with a linkage gripper remotely driven by a pneumatic rolling diaphragm actuator. The gripper is driven by underactuated mechanisms sensorized by joint encoders in order to fully determine the gripper state. We present the modelling of the linkage and fluidic transmission, validate its ability to regulate pinch force via admittance control within an RMS error well under 0.5 Newtons, and show the ability to detect contact at targeted locations on the linkage. In addition, we demonstrate simple grasping behaviors: blindly searching for an unobstructed object and detecting object loss. Our results show that an integrative approach of instrumenting underactuated gripper mechanisms can result in a lightweight gripper that is not only mechanically adaptive but sensitive enough to react to contact events without distal sensors or vision.

Automated grasping has a long history of research that is increasing due to interest from industry. One grand challenge for robotics is Universal Picking: the ability to robustly grasp a broad variety of objects in diverse environments for applications from warehouses to assembly lines to homes. Although many researchers now openly share code and data, it is challenging to compare and/or reproduce experimental results to identify which aspects of which approaches work best due to variations in assumptions and experimental protocols, e.g., sensors, lighting, robot arms, grippers, and objects.

While techniques exist for the rapid prototyping of mechanical and electrical components separately, this paper describes a method where commercial Additive Manufacturing (AM) techniques can be used to concurrently construct the mechanical structure and electronic circuits in a robotic or mechatronic system. The technique involves printing hollow channels within parts that are then filled with a low melting point liquid metal alloy that solidifies upon cooling to form electrical traces. This method is compatible with most conventional fused deposition modeling and stereolithography machines, and requires no modification to an existing printer, though the technique could easily be incorporated into multi-material machines. Three primary considerations are explored using the a commercial fused deposition manufacturing (FDM) process as a testbed: material and manufacturing process parameters, simplified injection fluid mechanics, and automatic part generation using standard printed circuit board software tools. As demonstration of the ability to embed circuit in RP parts, a differential-drive robot is printed, populated with discrete electronic components, and injected to create a fully functional robot.

This paper discusses the design of legged walking robots that are exactly constrained during the stance phase of locomotion. Legged robots with a large number of actuated degrees of freedom, while allowing for the widest range of controllable foot placements, often end up with overconstrained kinematics when in contact with the ground, requiring complex redundant control schemes for effective locomotion. Exactly-constrained robots would be capable of full body mobility while avoiding the weight and complexity costs of fully actuating each joint and would also allow for simpler control schemes. We discuss the constraints and degrees of freedom of a common legged robot kinematic structure and describe strategies for removing redundant constraints. Two major design considerations — architectural singularities and the uniqueness of the ground reaction forces — are discussed along with potential solutions. Finally, a prototype exactly-constrained walking robot is presented as a validation of this design strategy.

This paper examines aspects of robot hand performance specific to grasping and perching from an aerial vehicle and shows how various hand design parameters affect performance. Specifically, we consider hand performance when subject to external forces imparted to the hand from carrying a payload or from perching on a fixed item and explore the impact of design and grasp parameters including tendon routing/pulley ratio, object size, and palm size on the performance of both fully and underactuated designs. Our results show that underactuated designs utilizing a single actuator per finger are sufficient in all cases we studied, but that fully actuated designs can perform better for perching applications. Additionally, we find that increasing the palm width improves performance both when perching and grasping, and that a small distal/proximal pulley ratio is beneficial for payload carriage but counterproductive for perching.

Commercially available robotic hands are often expensive, customized for specific platforms, and difficult to modify. In this paper, we present the design of an open-source, low-cost, single actuator underactuated hand that can be created through fast and commonly-accessible rapid-prototyping techniques and simple, off-the-shelf components. This project establishes the design of an adaptive, four-finger hand utilizing simple 3D-printed components, compliant flexure joints, and readily obtainable off-the-shelf parts. Modular and adjustable finger designs are provided, giving the user a range of options depending on the intended use of the hand. The design tradeoffs and decisions made to achieve the 3D-printable, compact and lightweight robotic gripper are discussed, as well as a preliminary discussion of the performance differences between the finger designs. The authors intend this work to be the first in a series of open-source designs to be released, and through the contributions of the open-source user community, result in a large number of design modifications and variations available to researchers.

In this paper, we demonstrate an underactuated finger design and grasping method for precision grasping and manipulation of small objects. Taking inspiration from the human grasping strategy for picking up objects from a flat surface, we introduce the flip-and-pinch task, in which the hand picks up a thin object by flipping it into a stable configuration between two fingers. Despite the fact that finger motions are not fully constrained by the hand actuators, we demonstrate that the hand and fingers can interact with the table surface to produce a set of constraints that result in a repeatable quasi-static motion trajectory. Even when utilizing only open-loop kinematic playback, this approach is shown to be robust to variation in object size and hand position. Variation of up to 20° in orientation and 10 mm in hand height still result in experimental success rates of 80% or higher. These results suggest that the advantages of underactuated, adaptive robot hands can be carried over from basic grasping tasks to more dexterous tasks.

This video demonstrates the use of an underactuated robotic hand modified for the flip-and-pinch task to pick up thin objects from a table surface. Though well-suited for power-grasping, underactuated hands have difficulty with pinch-grasping and precision motions. We introduce a repeatable and robust method by which an underactuated hand flips thin objects off the table into a stable pinch grasp. We explain why this task is quasi-static and robust for a wide range of object dimensions.

In this paper we demonstrate an underactuated finger design and grasping method for precision grasping and manipulation of relatively small objects. Taking a cue from human manipulation, we introduce the flip-and-pinch task, in which the hand picks up thin objects from a table surface by flipping it into a stable configuration. Despite the fact that finger motions are not fully constrained by the hand actuators, we demonstrate that the hand and fingers can be configured with the table surface to produce a set of constraints that result in a repeatable quasi-static motion trajectory. This approach is shown to be robust for a variety of object sizes, even when utilizing identical open-loop kinematic playback. Experimental results suggest that the advantages of underactuated, adaptive robot hands can be carried over to dexterous, precision tasks as well.

This paper presents a concept for extending the functionality of robot hands so that they can better manipulate objects too small for an enveloping power grasp. Rather than pinching these objects between the fingertips of a hand, a miniature hand is embedded recursively on the end of a finger, enabling a power grasp on a smaller scale. The mechanics of designing such a gripper to operate without adding additional tendons are analyzed within the framework of underactuated elastic mechanisms. A simplified robot hand is demonstrated having a recursive gripper, and the process of picking up a pen and writing using a multi-scale grasp is demonstrated.

This paper presents a new method to produce computationally efficient models of robots that have planar elastic flexure joints. An accurate, low-dimensional model of large deformation bending is important to precisely describe the configuration of a flexure-jointed manipulator. The new model is based on the assumption that the curvature of a beam in bending is smooth and, thus, can be approximated by low-order polynomials. This produces a description of flexure motion that can be used as a joint model when expressed as a homogeneous transformation between rigid links--essentially a “drop in” replacement for traditional joint models such as screw coordinates and Denavit-Hartenberg conventions. Derivatives of the joint kinematics such as Jacobians and Hessians are accurate and easy to compute. We will show that with only three parameters, this model faithfully reproduces the elastic deformation of a flexure hinge predicted by the continuum model, even for large angles, without requiring numerical integration or many finite elements. The model can also be used to accurately compute the compliance and compressive buckling load of the flexure, as predicted by the continuum model.

In this paper we show that it is possible to design underactuated robot hands capable of performing dexterous manipulation tasks, despite the fact that the motion of an underactuated hand is not fully constrained by its actuators. If a robot has elastic elements at its joints, then the velocity of the actuators can be mapped onto the velocity of the grasped object using elastic averaging. This mapping can be used to compute classical measures of manipulability for an underactuated hand. We also demonstrate that holonomically constrained grasps can be analyzed to determine the manifold of stable object configurations that can be reached from some initial grasp. This is especially useful for planar manipulation operations, such as twisting a knob or precision positioning. A prototype two-fingered planar underactuated hand is introduced, having the ability to stably grasp and manipulate objects within the hand.

Sheet hinges, thin flexures that are rigid in the plane but which can bend freely, are common in stamped and lithographically manufactured devices. The behavior of these machine elements as joints in a robot is difficult to model because they are two-dimensional continuum elastic bodies that admit three-dimensional motion and twisting. This paper presents a parametric modeling technique that can be used to accurately predict elastic behavior of sheet hinges in three dimensions. Parameterized backbone curves can be used to represent ruled surface bending in a fashion that implicitly accounts for some of the complex boundary conditions imposed on typical sheet hinges. Approximate methods of integrating the non-commutative equations defining the sheet hinge backbone curves will be discussed, demonstrating acceptable trade-offs between accuracy and representational simplicity in overall model performance.

This paper presents new experimental results from a human-size robotic forearm, created to demonstrate the effectiveness of recruitment-based control architectures for large actuators made from shape memory alloys (SMA) and other active materials. The robot arm is actuated antagonistically by two actuators made up of 60 small SMA springs arranged in parallel, which are activated in an on/off fashion using Joule heating. The force and stiffness of each actuator is controlled by recruiting a desired number of springs to contract. The joint position is then controlled using equilibrium point servo control. The results presented in this paper show that the combination of equilibrium point control of the arm joint and recruitment-based control of each actuator's stiffness solve some of the major problems of scalability and response speed often associated with active material actuators.

This paper presents a control architecture for artificial muscle materials such as shape memory alloys and polymer actuators. The active material is broken up into many small independent cells that can be regulated in a binary fashion into ON and OFF states, so that the actuator displacement is determined by the number of ON cells. In order to control the number of cells that contract, a novel closed loop feedback control method is employed. Each cell is given a small stochastic finite state machine that governs its transition between ON and OFF states. A central controller globally varies the probabilistic rate with which all of the cells make state transitions. Using this architecture, actuator displacement can be controlled in a stable, robust fashion. Different feedback laws are compared using the fixed policy value iteration algorithm to calculate expected settling time in response to a step reference. A simple law based on calculating expected future behavior is found to be very close to the optimal law computed using the value iteration algorithm. The performance of the control laws is verified on a 50 cell shape memory alloy cellular actuator.

In this paper, the concept of broadcast feedback for stochastic cellular control systems is expanded to a system with nonuniform cellular length and nonuniform transition probability. The cellular control architecture was originally inspired by skeletal muscles comprising a vast number of tiny functional units, called sarcomeres. The output of the actuator system is an aggregate effect of numerous cellular units, each taking a bistable ON-OFF state. A central controller broadcasts the error between the aggregate output and a reference input. Rather than dictating the individual units to take specific states, the central controller merely broadcasts the overall error signal to all the cellular units uniformly. In turn each cellular unit makes a stochastic decision with a state transition probability, which is modulated in relation to the broadcasted error. Stability conditions of the broadcast feedback system are obtained by using a stochastic Lyapunov function. It is demonstrated that, even in the presence of the distribution of the cell length and/or the distribution of the transition probability generated in each cell, the aggregate output of the cellular units can track a given trajectory stably and robustly.

A broadcast-probability approach for the coordination of a vast number of actuators is proposed. We consider an actuator that consists of many "cellular" actuator units like human muscles. Each local unit has a decision-making unit which decides whether to accept or ignore the broadcasted control signal in a stochastic manner. We demonstrate through simulation that centralized coordination is unnecessary, and redundancy and randomness of the system improve the performance