Conventional dexterous grippers are not only heavy and complex but also task dependent. As a result, they cannot provide power, intermediate and precision grasping together. We propose usability evaluation of a novel gripper mechanism which can provide 1-Degree of Freedom (DOF) parallel grasping and 2-DOF independent pinching by using a closed chain architecture. The usability of our design and one commercial product has been evaluated based on various grasping tasks: adaptive grasping, parallel pinching grasping, independent pinching grasping and contact grasping. The success rate has been evaluated to validate the excellent usability of the proposed 3-DOF gripper.

This paper proposes a method for path planning of high-degree of freedom (DOF) articulated robots with adaptive dimensionality. For an efficient path planning in high-dimensional C-space (configuration space), first we describe an adaptive body selection that selects the robot bodies and joints depending on the complexity of path planning. It means that the robot may use necessary DOF to achieve a path planning task. The adaptive body selection method builds the C-space with adaptive dimensionality for a sampling-based path planner. Next, by using the adaptive body selection, the adaptive Rapidly-Exploring Random Tree (RRT) algorithm is introduced, which incrementally grows RRTs in the adaptive dimensional C-space. And we show through several simulation results that the proposed method is more efficient than the original RRT-based path planner, which requires full-dimensional planning.

This study developed the Hanyang Exoskeleton Assistive Robot (HEXAR)-CR50 aimed at improving muscle strength of the wearer while transporting a load. The developed exoskeleton robot HEXAR-CR50 has 7 degrees of freedom (DOF) for one foot, 3-DOF for the hip joints, 1-DOF for the knee joints, and 3-DOF for the ankle joints. Through functional analysis of each human joint, two DOFs were composed of active joints using an electric motor developed in an under-actuated form with heightened efficiency. The rest of the DOFs were composed of passive or quasi-passive joints to imitate human joints. The control of the exoskeleton robot was based on the physical human-robot interaction. In order to verify the performance of the developed HEXAR-CR50, muscle activity was measured using electromyography, vGRF was measured using F-Scan sensor. The experimental results showed that %MVIC was reduced against the external load applied, while GRF had a decrement rate, compared with the external load when the exoskeleton was worn, which verified the performance in accordance with the development objective of load carrying. A muscle strength augment effect from the developed wearable robot was verified.

In this paper, we propose and experimentally test a dynamic model-based force controller for the motion of upper limb exoskeleton robot. The system is composed of 3 degrees of freedom using an electrical actuator. This system is mainly controlled by the multi-axis force sensor signals. These are used to generate desired torques for driving the robot system. However, singularities exist when force signals in the Cartesian coordinate system are transformed to torques in the joint coordinate system. So we applied the damped least squares method. In handling loads, torque compensation regarding the weight of the object is required. Therefore, we installed the multi-axis force sensor at the robot end effector. It measures the interaction forces between the exoskeleton and the load. To compensate for the handling object, we used the static model. We performed control stability and load handling experiments to verify the effectiveness of the controller. With these experiments, we confirmed the effectiveness of the proposed controller.

This paper describes the compact design of a torque sensor using optical technique and its fabrication for wearable and quadruped robots. The torque generated can only be measured by placing torque sensor on each joint, which makes the joint bulky and heavy, adding extra load to be carried by robot. In this paper we proposed a compact and light weight design for the torque sensor. This sensor can measure torque of robots links by detecting torsion. The optical technique is used to detect angular displacement in the joints. The torque can be calculated by using torque and twist-angle relation. The proposed design is simulated by FEM software (ANSYS) and it shows successful measurements of the torque with a load capacity of 50 Nm, which is sufficient for the torque generated in wearable and quadruped robots joints. The designed optical torque sensor is manufactured by EDM (Electrical Discharge Machining). The optical torque sensor was calibrated and several experiments were conducted to ratify its feasibility with the wearable and quadruped robots.

Coverage path planning determines a path that passes a robot, a detector, or some type of effector over all points in the environment. Prior work in coverage tends to fall into one of two extremes: coverage with an effector the same size of the robot, and coverage with an effector that has infinite range. In this paper, we consider coverage in the middle of this spectrum: coverage with a detector range that goes beyond the robot, and yet is still finite in range. We achieve coverage in two steps: The first step considers vast, open spaces, where the robot can use the full range of its detector; the robot covers these spaces as if it were as big as its detector range. Here we employ previous work in using Morse cell decompositions to cover unknown spaces. A cell in this decomposition can be covered via simple back-and-forth motions, and coverage of the vast space is then reduced to ensuring that the robot visits each cell in the vast space. The second step considers the narrow or cluttered spaces where obstacles lie within detector range, and thus the detector "fills" the surrounding area. In this case, the robot can cover the cluttered space by simply following the generalized Voronoi diagram (GVD) of that space. In this paper, we introduce a hierarchical decomposition that combines the Morse decompositions and the GVDs to ensure that the robot indeed visits all vast, open, as well as narrow, cluttered, spaces. We show how to construct this decomposition online with sensor data that is accumulated while the robot enters the environment for the first time.

This paper presents a method for a planar rigid body consisting of multiple convex bodies to explore an unknown planar workspace, i.e., an unknown configuration space diffeomorphic to SE(2). This method is based on a roadmap termed concave hierarchical generalized Voronoi graph (concave-HGVG). Just as in our previous work, we decompose the free configuration space into contractible cells in which we define the concave generalized Voronoi graphs (concave-GVG), and then connect these graphs using an additional structure termed one-tangent edges. Since the robot consists of multiple convex bodies, the one-tangent edges are defined using the diameter function of the convex hull of the convex bodies as well as the individual convex bodies. These two structures together form the concave-HGVG, which is a one-dimensional roadmap of the multi-convex bodies in plane. Both components are defined in terms of workspace distance measurement, and thus the concave-HGVG can readily be constructed in a sensor-based way.

We present a new roadmap that can be used to guide a convex body to explore an unknown planar workspace, i.e., to map an unknown configuration space diffeomorphic to SE(2). This new roadmap is called the convex hierarchical generalized Voronoi graph (convex-HGVG). Since this roadmap is defined in terms of workspace distance information that is within line of sight of the convex body, we can use it to direct the robot to explore an unknown configuration space diffeomorphic to SE(2). The challenge in defining the roadmap is that SE(2), with holes removed from it, generally does not have a one-dimensional deformation retract. Therefore, we decompose the punctured SE (2) into contractible regions, in which we define convex generalized Voronoi graphs (convex-GVG), and then connect these graphs with additional structures called convex-R edges. We formally show that the convex-HGVG, which is the union of the convex-GVG edges and the convex-R edges, is indeed a roadmap.

This work considers the motion planning for a rod-shaped robot in a three-dimensional space. Our approach is to construct a roadmap in the configuration space of the rod and use it to navigate and map unknown spaces. Previously, we defined the rod-HGVG, which is a roadmap for a rod-shaped robot operating in a three-dimensional space. However, the rod-HGVG is only guaranteed to be connected if the point-GVG is connected. This work extends the rod-HGVG into more general environments where the point-GVG may not be connected but the point-HGVG is connected. For this, we define new components, termed the higher-order edges for the rod-HGVG and linking strategies which are based on the point-HGVG. Just as there is a close relationship between the components of the point-GVG and the rod-HGVG, there is a close relationship between the components of the point-HGVG and the higher-order edges of the rod-HGVG. We provide the construction procedures for the higher-order edges, which, like the components of the rod-HGVG, can be constructed using only sensor-provided information.

We present an algorithm that enables a convex shaped robot to explore an unknown planar workspace, i.e., an unknown configuration space diffeomorphic to SE(2). This algorithm is based on a roadmap called the convex hierarchical generalized Voronoi graph (convex-HGVG), which is the central contribution of the paper. Since this roadmap is defined in terms of workspace distance information which is within line of sight of the convex body, we can use this roadmap to direct the robot to explore the unknown space. The challenge in defining the roadmap is that SE(2), with holes removed from it, does not have, in general, a one-dimensional retract. Therefore, we decompose the punctured SE(2) into contractible regions in which we define convex generalized Voronoi graphs (convex-GVG) and then connect these graphs with additional structures called R-edges. The convex-GVG edges and the R-edges together form the convex-HGVG. We show that the convex-HGVG is indeed a roadmap and then we demonstrate the control laws that a robot can invoke that incrementally constructs the roadmap.

We describe a new roadmap, termed the rod-HGVG, for motion planning of a rod-shaped robot operating in a three-dimensional volume. This roadmap is defined in terms of work space distance enabling us to prescribe an incremental construction procedure. This allows the rod to explore its configuration space, R/sup 3//spl times/S/sup 2/, without ever explicitly constructing the configuration space. In fact, the rod robot need not know the work space ahead of time. We term the rod-HGVG a piecewise retract because it comprises many retracts. Homotopy theory asserts that there cannot be in general a one dimensional retract of noncontractable five-dimensional space. Instead, we define an exact cellular decomposition on R/sup 3//spl times/S/sup 2/ and a retract in each cell. Next, we "connect" the retracts of each cell forming a piecewise retract of the rod's configuration space.

Sensor-based planning for rod-shaped robots is necessary for the realistic deployment of noncircularly symmetric robots into unknown environments. Whereas circularly symmetric robots have two-dimensional Euclidean configuration spaces, planar rod robots posses three degrees-of-freedom, two for position and one for orientation, and hence have a three-dimensional configuration space, SE(2). In this work, we define the rod hierarchical generalized Voronoi graph (rod-HGVG) which is a roadmap of the rod's configuration space. Prior work in Voronoi-based roadmaps use a retraction of the robot's free space to define the roadmap; here, we break apart the robot's free space into regions where fragments of the roadmap are defined and then connect the fragments. The primary advantage of the rod-HGVG is that it is defined in terms of workspace distance measurements, which makes it amenable to sensor-based planning. This paper also includes a numerical procedure that generates the rod-HGVG edge fragments using only information that is within line of sight of the rod robot. It is worth noting that this procedure does not require an explicit definition of configuration space, i.e., this procedure constructs a roadmap of rod configuration space without ever constructing the configuration space itself.