Tangential motion on a finger pad is a promising method of transmitting directional tactile information to human users. This study examined the identification and discrimination of tangential force motion on an index finger pad. An experimental device was built to automatically and randomly move a small probe in eight radial directions (45° apart) and two distances (0.5 and 1.5 mm). Index fingers of 62 subjects were tested. The results showed that moving the probe at 1.5 mm was detected with more accuracy than the 0.5 mm one. And, the absolute direction was not a statistically significant variable affecting accuracy for 1.5 mm distance, but was a significant effect for 0.5 mm distance. Implications of these results are discussed and future developments are offered within the context of a proposed Braille design with tangential actuators.

This paper presents a new approach to matching occupancy grid maps by means of finding correspondences between a set of sparse features detected in the maps. The problem is stated here as a special instance of generic image registration. To cope with the uncertainty and ambiguity that arise from matching grid maps, we introduce a modified RANSAC algorithm which searches for a dynamic number of internally consistent subsets of feature pairings from which to compute hypotheses about the translation and rotation between the maps. By providing a (possibly multi-modal) probability distribution of the relative pose of the maps, our method can be seamlessly integrated into large-scale mapping frameworks for mobile robots. This paper provides a benchmarking of different detectors and descriptors, along extensive experimental results that illustrate the robustness of the algorithm with a 97% success ratio in loop-closure detection for ~1700 matchings between local maps obtained from four publicly available datasets.

A small-scale experimental setup for autonomous target tracking of a surface vessel in the presence of obstacles is presented. The experiments are performed in simulated rough seas through wave, current, and wind generation in a small indoor pool. Absolute position of the agent and the target as well as the obstacle size and position are provided through an overhead camera by detecting color light emitting diodes installed on all objects. Ordinary differential equations with stable limit-cycle solutions are used to define transitional trajectories around obstacles based on the camera data. A sliding mode control law is implemented for real-time tracking control which is capable of rejecting large disturbances from the generated waves and wind. The sliding mode control signals are sent to wireless receivers on the autonomous vessel where a proportional integral speed controller maintains the commanded speed. A special scaling method is presented to show that the environmental forces are similar to those of moderate through high sea states. Several experiments are presented where the autonomous vessel catches and follows a target boat moving in arbitrary trajectories in both the presence and absence of obstacles.

This paper proposes a repetitive control type optimal gait generation framework by executing learning control and parameter tuning. We propose a learning optimal control method of Hamiltonian systems unifying iterative learning control (ILC) and iterative feedback tuning (IFT). It allows one to simultaneously obtain an optimal feedforward input and tuning parameter for a plant system, which minimizes a given cost function. In the proposed method, a virtual constraint by a potential energy prevents a biped robot from falling. The strength of the constraint is automatically mitigated by the IFT part of the proposed method, according to the progress of trajectory learning by the ILC part.

It has been shown before that the Rapidly Exploring Random Tree (RRT) algorithm is probabilistically and resolution complete; and that the probability of finding a particular solution path can be related to the number of nodes. However, little analysis exists on the rate at which the tree covers the configuration space. In this paper, we present a stochastic difference equation which models how the tree covers the configuration space as a function of the number of nodes in the tree. Using two simplifying assumptions, appropriate for holonomic, kinematic systems in expansive configuration spaces, we derive closed-form solutions for the expected value and variance of configuration space coverage, which only depend on two easily computable parameters. Using a grid-based coverage measurement, we present experimental evidence supporting this model across a range of dimensions, obstacle densities, and parameter choices. Collecting data from 1000 RRTs, we provide evidence that configuration space coverage concentrates tightly around the expected coverage predicted by the model; and the results of the Chi-squared test suggest that the distribution of coverage across these runs is highly Gaussian. Together these results enable one to predict the expected coverage, along with a confidence interval, after a certain number of nodes have been added to the tree. We also applied the model to an example with extremely narrow passages and to a system with non-holonomic kinematics. The expected value prediction is still qualitatively accurate; but the rate constant is reduced and the variance is higher. Overall, in addition to its theoretical value, the model may find future application as an online measure of search-progress and problem difficulty, useful for adaptive variants of the basic RRT algorithm.

We consider the conventional techniques of vision robot system calibration where camera parameters and robot hand–eye parameters are computed separately, i.e., first performing camera calibration and then carrying out hand–eye calibration based on the calibrated parameters of cameras. In this paper we propose a joint algorithm that combines the camera calibration and the hand–eye calibration together. The proposed algorithm gives the solutions of the cameras' parameters and the hand–eye parameters simultaneously by using nonlinear optimization. Both simulations and real experiments show the superiority of our algorithm. We also apply our algorithm in the real application of the robot-assisted surgical system, and very good results have been obtained.

Clinical outcomes have shown that robot-assisted rehabilitation is potential of enhancing quantification of therapeutic process for patients with stroke. During robotic rehabilitation exercise, the assistive robot must guarantee subject's safety in emergency situations, e.g., sudden spasm or twitch, abruptly severe tremor, etc. This paper presents a hierarchical control strategy, which is proposed to improve the safety and robustness of the rehabilitation system. The proposed hierarchical architecture is composed of two main components: a high-level safety supervisory controller (SSC) and low-level position-based impedance controller (PBIC). The high-level SSC is used to automatically regulate the desired force for a reasonable disturbance or timely put the emergency mode into service according to the evaluated physical state of training impaired limb (PSTIL) to achieve safety and robustness. The low-level PBIC is implemented to achieve compliance between the robotic end-effector and the impaired limb during the robot-assisted rehabilitation training. The results of preliminary experiments demonstrate the effectiveness and potentiality of the proposed method for achieving safety and robustness of the rehabilitation robot.

We consider the connectivity of autonomous mobile robots. The robots navigate using simple local steering rules without requiring explicit communication among themselves. We show that using only position information of neighbors, the group connectivity can be sustained even in the case of bounded position measurement errors and the occlusion of robots by other robots in the group. In implementing the proposed scheme, sub-optimal solutions are invoked to avoid an excessive computational burden. We also discuss the possibility of deadlock which may bring the group to a standstill and show that the proposed methodology avoids such a scenario in real-life settings.

In most of our daily motion tasks, the coordination between limbs is very crucial for successful execution of the tasks. In this paper, coordination among oscillators controlling in Cartesian space is studied to control bipedal walking. In our method, phase adjustment among oscillators is considered as one of the key issues to achieve coordination. A new phase adjustment method is proposed. With this method, an oscillator is able to coordinate other oscillators and maintain a desired phase relationship. This property is important for the walking control especially when external perturbations are given. To simplify the relationship between oscillators in a central pattern generator (CPG), a hierarchical CPG structure is adopted, where a main oscillator will be used to adjust other oscillators. In the simulation, the walking motion controlled by the CPG controller converges to a stable pattern even with external perturbations. We have implemented the controller in both the simulation model and real hardware robot.

The majority of current robotic joints are primarily actuated by rotational mechanisms. These electrical drives have substantially different features than the features found in human muscular systems. This paper presents a cost-effective solution to the backlash of a phenomenon known to cause positioning errors and other undesirable dynamic effects in drives. These errors are particularly pronounced when relatively major changes appear in the pre-load conditions of the motor such as in the case of a robotic leg or arm with a high degree of freedom. Current solutions require an accurate time-varying model of drives that is not available in the majority of practical cases. Therefore, in this paper a digitally controlled mechanical solution is proposed which is inspired by the human flexor–extensor mechanism. The idea is to construct an antagonistic actuator pair analogous to the flexor and extensor muscles. In order to obtain good control performance even in the low-speed range, permanent magnet stepper motors were chosen as actuators that are commutated in a digitally closed-loop fashion. The operation of the controlled structure has been verified in a real experimental environment where measurements showed good results and match with previous simulations.

The combined motion of the human thumb, index and middle fingers while rotating a small object across the extended, intermediate and flexed planes with respect to the fingers was analyzed. Auto reflective markers were attached on the fingers to track their motion across three postures and planes via a 3D motion capture system. Central, right and left rotation postures were considered in each plane for investigation and the rotation experiments were performed with 30 healthy subjects. The obtained data were used to compute the finger joint angles. Based on the three criteria of (i) manipulability measure, (ii) major axis direction angle of the manipulability ellipsoid and (iii) ratio of the minor over major axis lengths, the collective behavior of the fingers was studied. It has been found after analysis that the thumb and middle finger were active, while the index finger operated passively when manipulating small objects in cooperative rotational motion across the three planes. Activeness refers to the independence of a digit in controlling the motion of an object whereas passiveness denotes its dependence on other digits. An active finger governs the motion of an object whereas a passive finger simply supports it. The results of this investigation are of great importance in planning treatment for rehabilitation and for designing controllers for robotic therapists, finger exoskeletons and prostheses.

This paper proposes a new visual control approach based on sinusoidal inputs to be used on a nonholonomic robot. We present several contributions: In our method, developed considering a unicycle kinematic model, sinusoids are used in such a way that the generated vehicle trajectories are feasible, smooth and versatile. Our technique improves previous sinusoidal-based control works in terms of efficiency and flexibility. As further contributions, we present analytical expressions for the evolution of the robot's state, and propose a new state-feedback control law based on these expressions. All the information used in the control scheme is obtained from omnidirectional vision by means of the one-dimensional trifocal tensor. Stability analysis of the proposed approach is presented, and its performance is illustrated through experiments.

In this paper, the force control of a constrained one-link flexible arm is investigated using a feedback parallel compensation algorithm based on a linear distributed parameter model with internal damping of Kelvin–Voigt type. Generally, the non-collocation of the joint torque input and the tip contact force output comes along with the non-minimum phase in nature. To overcome this inherent limitation, a new input induced by the measurement of root-bending moment and its derivative, and a virtual contact force output generated by a parallel compensator are defined. Therefore, the transfer function from the new input to the virtual contact force output is proved not only strictly minimum phase but also in a stable condition. A PD controller then improves the performance of the overall closed-loop system. Furthermore, the perfect asymptotic tracking of a desired contact force trajectory with internal stability can be achieved accurately. The exact solutions of the infinite-dimensional system are obtained using the infinite product formulation. The proposed system promises stability robustness to parameter uncertainties, also free of spillover problems. Numerical simulations are provided to verify the effectiveness of the proposed approach.