ICRA 2020 Papers

Soft active materials can generate flexible locomotion and change configurations through large deformations when subjected to an external environmental stimulus. They can be engineered to design 'soft machines' such as soft robots, compliant actuators, flexible electronics, or bionic medical devices. By embedding ferromagnetic particles into soft elastomer matrix, the ferromagnetic soft matter can generate flexible movement and shift morphology in response to the external magnetic field. By taking advantage of this physical property, soft active structures undergoing desired motions can be generated by tailoring the layouts of the ferromagnetic soft elastomers. Structural topology optimization has emerged as an attractive tool to achieve innovative structures by optimizing the material layout within a design domain, and it can be utilized to architect ferromagnetic soft active structures. In this paper, the level-set-based topology optimization method is employed to design ferromagnetic soft robots (FerroSoRo). The objective function comprises a sub-objective function for the kinematics requirement and a sub-objective function for minimum compliance. Shape sensitivity analysis is derived using the material time derivative and adjoint variable method. Three examples, including a gripper, an actuator, and a flytrap structure, are studied to demonstrate the effectiveness of the proposed framework.

Robot rehabilitation is an emerging and promising topic that incorporates robotics with neuroscience and rehabilitation to define new methods for supporting patients with neurological diseases. As a consequence, the rehabilitation process could increase the efficacy exploiting the potentialities of robot-mediated therapies. Nevertheless, nowadays clinical effectiveness is not enough to widely introduce robotic technologies in such social contexts. In this paper we propose a step further, presenting an innovative exoskeleton for wrist flexion/extension and adduction/abduction motion training. It is designed to be wearable and easy to control and manage. It can be used by the patient in collaboration with the therapist or autonomously. The paper introduces the main steps of device design and development and presents some tests conducted with an user with limited wrist mobility.

This paper describes the development of a novel non-anthropomorphic robot hand with the ability to manipulate objects by means of articulated, actively driven rollers located at the fingertips. An analysis is conducted and systems of equations for two-finger and three-finger manipulation of a sphere are formulated to demonstrate full six degree of freedom nonholonomic spatial motion capability. A prototype version of the hand was constructed and used to grasp and manipulate a variety of objects. Tests conducted with the prototype confirmed the validity of the mathematical analysis. Unlike conventional approaches to within-hand manipulation using legacy robotic hands, the continuous rotation capability of our rolling fingertips allows for unbounded rotation of a grasped object without the need for finger gaiting.

This paper proposes a 6 degree of freedom (DoF) manipulator for haptic application. The proposed haptic device, named GHap, is designed based on the four-bar-linkage mechanism for linear motion with the ring-type gimbal mechanism. To improve the force display ability, the device is designed to compensate the gravity force of the manipulator by its own weight. The conceptual mechanical design is compared by placing the third joint, which controls the four-bar mechanism, in two different configurations. The forward kinematics and the jacobian of GHap are presented. Finally, the gravity compensation method and open-loop force display performance of the proposed haptic device are validated by an experiment with the GHap prototype.

This paper identifies conditions for designing the appropriate spatial admittance to achieve reliable force-guided assembly of polyhedral parts for cases in which a single frictional contact occurs between the two parts. This work is an extension of previous work in which frictionless contact was considered. This paper presents a way to characterize friction without solving a set of complicated non-linear equations. We show that, by modifying the error reduction function and evaluating the function bounds associated with friction, the procedures developed for frictionless contact apply to the frictional cases. Thus, for bounded misalignments, if an admittance satisfies the misalignment-reducing conditions at a finite number of contact configurations, then the admittance will also satisfy the conditions at all intermediate configurations for any value of friction less than the specified upper bound.

Recently, it has become possible to easily design and fabricate robotic systems in the laboratory and at home due to the recent development of 3D printer technology. On the other hand, the strength of the plastic materials used in reasonably priced 3D printers and the accuracy of the printed parts are generally low. These problems affect the part-joining quality. Therefore, this paper describes a design method inspired by ancient Japanese wooden joinery techniques for assembling 3D-printed parts and presents the design of a robotic hand as its application. The joinery techniques use special shapes to assemble components and allow us to assemble the robotic hand without glue, screws or nails and to easily disassemble it.

For the majority of tasks performed by traditional serial robot arms, such as bin picking or pick and place, only two or three degrees of freedom (DOF) are required for motion; however, by augmenting the number of degrees of freedom, further dexterity of robot arms for multiple tasks can be achieved. Instead of increasing the number of joints of a robot to improve flexibility and adaptation, which increases control complexity, weight, and cost of the overall system, malleable robots utilise a variable stiffness link between joints allowing the relative positioning of the revolute pairs at each end of the link to vary, thus enabling a low DOF serial robot to adapt across tasks by varying its workspace. In this paper, we present the design and prototyping of a 2-DOF malleable robot, calculate the general equation of its workspace using a parameterisation based on distance geometry-suitable for robot arms of variable topology, and characterise the workspace categories that the end effector of the robot can trace via reconfiguration. Through the design and construction of the malleable robot we explore design considerations, and demonstrate the viability of the overall concept. By using motion tracking on the physical robot, we show examples of the infinite number of workspaces that the introduced 2-DOF malleable robot can achieve.

This paper presents a soft robotic ankle-foot orthosis (SR-AFO) exosuit designed to provide support to the human ankle in the frontal plane without restricting natural motion in the sagittal plane. The SR-AFO exosuit incorporates inflatable fabric-based actuators with a hollow cylinder design which requires less volume than the commonly used solid cylinder design for the same deflection. The actuators were modeled and characterized using finite element analysis techniques and experimentally validated. The SR-AFO exosuit was evaluated on healthy participants in both a sitting position using a wearable ankle robot and a standing position using a dual-axis robotic platform to characterize the effect of the exosuit on the change of 2D ankle stiffness in the sagittal and frontal planes. For both sitting and standing test protocols, a trend of increasing ankle stiffness in the frontal plane was observed up to 50 kPa while stiffness in the sagittal plane remained relatively constant over pressure levels. During quiet standing, the exosuit could effectively change eversion stiffness at the ankle joint from about 20 to 70 Nm/rad at relatively low- pressure levels (<; 30 kPa). Eversion stiffness was 84.9 Nm/rad at 50 kPa, an increase of 387.5% from the original free foot stiffness.

Needle-based procedures, such as biopsy and percutaneous tumor ablation, highly depend on the accuracy of needle placement. The accuracy is significantly affected by the needle-tissue interaction no matter what needles (straight or steerable) are used. Due to the unknown tissue mechanics, it is challenging to achieve high accuracy in practice. This paper hence proposes a needle design with an articulated tip for increased steerability and improved needle path consistency. Due to the passive needle tip articulation, tissue mechanics always plays a dominant role such that the needle creates similar paths with approximately piece-wise constant curvature in different tissues. Kinematics model for the proposed needle is presented. The algorithms of path planning and needle tip pose estimation under external imaging modality are developed. Experimental verifications were conducted to demonstrate the needle's steerability as well as the target-reaching capability with obstacles avoidance.

This paper introduces a new hydraulic manipulator with hydraulic-cable driven actuation (HCA) modules for disaster response mobile-manipulation. The hydraulic actuation system has the potential to apply disaster-response application, because it has a higher power-to-weight ratio and robustness to external impacts than electric motor actuation. However, using a conventional hydraulic manipulators is inappropriate because the revolute joint uses conventional actuators, such as linear cylinders and vanes, which have some limitations: 1) linear cylinder: small range of motion, 2) vane: low torque-toweight ratio. To overcome these limitations, we propose new 3DOF manipulator which has a larger workspace than the conventional hydraulic manipulator and comparable payloadto-weight ratio. To this end, we use hydraulic-cable driven actuation modules from our previous research. Experimental results verify the basic performance of the actuator modules and manipulator and their capability to perform various disaster response tasks.

Active force sensors are key instruments to get around the tradeoff between the sensitivity and the measurement range of conventional passive force sensors. Thanks to their quasi-infinite stiffness in closed loop, active sensors can be applied for force measurements on samples with a wide range of stiffness without interference with the mechanical parameters of the sensor. MEMS (Micro-Electro Mechanical Systems) active force sensors have been wildly developed in the literature but they are ill adapted for force measurements at the Newton level needed in meso-scale robotics. In this article, a novel structure for a meso-scale active force sensor is proposed for the measurement of forces from the milli-newton to the newton.This novel meso-scale sensor is based on a nil-stiffness guidance and an electromagnetic actuation. This paper deals with its design, identification, calibration and closed loop control. The sensor exhibits nil-stiffness characteristic in open loop and an almost infinite stiffness in closed loop. This allows measuring forces with a large range of gradients. First experiments shows the ability of this new sensor architecture to measure low frequency forces up to 0.8N with a precision of 0.03 N and a closed loop -20 dB cutoff frequency of 73.9Hz.

Snake-like robots for endoscopic surgery make it possible to reach deep-seated lesions. With the use of small flexible tendon-driven instruments, it is possible to perform bimanual micro-surgical tasks that are challenging for standard endoscopic surgeries. Existing devices, however, lack articulated wrists and rolling motion of the end-effector. This paper presents a new instrument design with a distal-roll gripper for snake-like robots. The developed 5 DoFs miniaturized instruments with a diameter of 3 mm enable the deployment into narrow endoluminal channels. Issues related to actuation coupling, tendon slack, and backlash are addressed. Experimental results show that the distal-roll gripper can rotate 106°, and the actuated joints can achieve good repeatability and accuracy with the proposed compensation control scheme.

Aircraft that can launch ballistically and convert to autonomous, free-flying drones have applications in many areas such as emergency response, defense, and space exploration, where they can gather critical situational data using onboard sensors. This paper presents a ballistically-launched, autonomously-stabilizing multirotor prototype (SQUID - Streamlined Quick Unfolding Investigation Drone) with an onboard sensor suite, autonomy pipeline, and passive aerodynamic stability. We demonstrate autonomous transition from passive to vision-based, active stabilization, confirming the multirotor's ability to autonomously stabilize after a ballistic launch in a GPS-denied environment.

Hand synergy from neuroscience provides an effective tool for anthropomorphic hands to realize versatile grasping with simple planning and control. This paper aims to extend the synergy-inspired design from anthropomorphic hands to multi-fingered robot hands. The synergy-inspired hands are not necessarily humanoid in morphology but perform primary characteristics and functions similar to the human hand. At first, the biomechanics of hand synergy is investigated. Three biomechanical characteristics of the human hand synergy are explored as a basis for the mechanical simplification of the robot hands. Secondly, according to the synergy characteristics, a three-fingered hand is designed, and its kinematic model is developed for the analysis of some typical grasping and manipulation functions. Finally, a prototype is developed and preliminary grasping experiments validate the effectiveness of the design and analysis.

Accurate depth estimation remains an open problem for robotic manipulation; even state of the art techniques including structured light and LiDAR sensors fail on reflective or transparent surfaces. We address this problem by training a neural network model to estimate depth from RGB-D images, using labels from physical interactions between a robot and its environment. Our network predicts, for each pixel in an input image, the z position that a robot's end effector would reach if it attempted to grasp or poke at the corresponding position. Given an autonomous grasping policy, our approach is self-supervised as end effector position labels can be recovered through forward kinematics, without human annotation. Although gathering such physical interaction data is expensive, it is necessary for training and routine operation of state of the art manipulation systems. Therefore, this depth estimator comes for free while collecting data for other tasks (e.g., grasping, pushing, placing). We show our approach achieves significantly lower root mean squared error than traditional structured light sensors and unsupervised deep learning methods on dif cult, industry-scale jumbled bin datasets.

We present DenseCAvoid, a novel algorithm for navigating a robot through dense crowds and avoiding collisions by anticipating pedestrian behaviors. Our formulation uses visual sensors and a pedestrian trajectory prediction algorithm to track pedestrians in a set of input frames and compute bounding boxes that extrapolate to the pedestrian positions in a future time. Our hybrid approach combines this trajectory prediction with a Deep Reinforcement Learning-based collision avoidance method to train a policy to generate smoother, safer, and more robust trajectories during run-time. We train our policy in realistic 3-D simulations of static and dynamic scenarios with multiple pedestrians. In practice, our hybrid approach generalizes well to unseen, real-world scenarios and can navigate a robot through dense crowds (~1-2 humans per square meter) in indoor scenarios, including narrow corridors and lobbies. As compared to cases where prediction was not used, we observe that our method reduces the occurrence of the robot freezing in a crowd by up to 48%, and performs comparably with respect to trajectory lengths and mean arrival times to goal.

Robot networks are susceptible to fail under the presence of malicious or defective robots. Resilient networks in the literature require high connectivity and large communication ranges, leading to high energy consumption in the communication network. This paper presents robot formations with guaranteed resiliency that use smaller communication ranges than previous results in the literature. The formations can be built on triangular and square lattices in the plane, and cubic lattices in the three-dimensional space. We support our theoretical framework with simulations.

Task-oriented dialogue system has a vital role in Human-Robot Interaction (HRI). However, it has been developed based on conventional pipeline approach which has several drawbacks; expensive, time-consuming, and so on. Based on this approach, developers manually define a robot's behaviour such as gestures and facial expressions on the corresponding dialogue states. Recently, end-to-end learning of Recurrent Neural Networks (RNNs) is an attractive solution for the dialogue system. In this paper, we proposed a social robot system using end-to-end dialogue system in the context of hospital receptionist. We utilized Hybrid Code Network (HCN) as an end-to-end dialogue system and extended to select both response and gesture using RNN based gesture selector. We evaluate its performance with human users and compare the results with one of the conventional methods. Empirical result shows that the proposed method has benefits in terms of dialogue efficiency, which indicates how efficient users were in performing the given tasks with the help of the robot. Moreover, we achieved the same performance regarding the robot's gesture with the proposed method compared to manually defined gestures.

Ocean Worlds represent one of the best chances for extra-terrestrial life in our solar system. A new mission concept must be developed to explore these oceans. This mission would require traversing the 10s of km thick icy shell and releasing a submersible into the ocean below. During the transit of the icy shell and the exploration of the ocean, the vehicle(s) would be out of contact with Earth for weeks or potentially months at a time. During this time the vehicle must have sufficient autonomy to locate and study scientific targets of interest. One such target of interest is hydrothermal venting. We have previously developed an autonomous nested search method to locate and investigate sources of hydrothermal venting by locating local maxima in hydrothermal vent emissions. In this work we demonstrate this approach on board an OceanServer Iver2 AUV in Chesapeake Bay, MD using simulated sensor data from a hydrothermal plume model. This represents the first step towards the deployment of this approach in conditions analogous to those that we might expect on an Ocean World.