Advanced Robotics

Savino, Sergio

doi: 10.1080/01691864.2013.804156pmid: N/A

This paper presents an algorithm for using the stereoscopic vision in environmental recognition and in particular in the motion detection. The algorithm is based on the recognition of certain environmental characteristics, in particular straight-line segments, to be identified in two distinct stereoscopic observations of the working environment. The knowledge of some environmental characteristics of a scene observed from different positions allows estimating the transformation between the observation positions. The study of such algorithm aims to promote the integration of stereoscopic vision in robotics, especially in mobile and autonomous robots, and to promote the use of these methodologies to improve the possibilities of control in the robotic applications. Some results are presented with reference to a stereoscopic vision system applied to a 3 d.o.f. revolute robot.

Petrič, Tadej; Gams, Andrej; Debevec, Tadej; Žlajpah, Leon; Babič, Jan

doi: 10.1080/01691864.2013.804164pmid: N/A

In this paper we compare three noninvasive control methods for a robotic knee exoskeleton and asses their kinematic influences on the repetitive squatting motions of able-bodied human subjects. The motion of the subjects wearing the knee exoskeleton was also compared to the motion of the subjects performing the same task without using the assistance of the knee exoskeleton. We chose the squatting motion because it approximates common movements with high metabolic cost, such as standing up from a chair and ascending or descending the stairs. Beside the two classical robotic control approaches, i.e. the position control and the gravity compensation, we propose a method that is based on a single adaptive frequency oscillator combined with an adaptive Fourier series in a feedback loop. The method can extract frequency and phase of an arbitrary periodic signal in real-time. This method is particularly appropriate for controlling novel robotic assisting devices since it does not require complex signal sensing or user calibration. The results show that the total knee torque was increased while using the exoskeleton device compared to the squatting without the assistance of the exoskeleton device. In effect, there were significant kinematic adaptations observed when the exoskleton device assisted the motion of the subjects. However, no significant kinematic differences were found between different control methods. We conclude that an assistive device can augment the abilities of the able-bodied humans in the targeted joints (i.e. the joints provided with additional mechanical power) but, on the other hand, significantly alters whole body kinematics.

Peternel, Luka; Babič, Jan

doi: 10.1080/01691864.2013.808305pmid: N/A

We present a novel approach where a human demonstrator can intuitively teach robot full-body skills. The aim of this approach is to exploit human sensorimotor ability to learn how to operate a humanoid robot in real time to perform tasks involving interaction with the environment. The human skill is then used to design a controller to autonomously control the robot. To provide the demonstrator with the robot’s state suitable for the full-body motion control, we developed a novel method that transforms robot’s sensory readings into feedback appropriate for the human. This method was implemented through a haptic interface that was designed to exert forces on the demonstrator’s centre of mass corresponding to the state of the robot’s centre of mass. To evaluate the feasibility of this approach, we performed an experiment where the human demonstrator taught the robot how to compliantly interact with another human. The results of the experiment showed that the proposed approach allowed the human to intuitively teach the robot how to compliantly interact with a human.

Dimeas, Fotios; Koustoumpardis, Panagiotis; Aspragathos, Nikos

doi: 10.1080/01691864.2013.804801pmid: N/A

In this paper, two admittance-based control schemes for a power-assisted lifting device are presented. This device can be used to hoist a heavy object interactively for reducing the operator’s burden. The proposed system integrates an admittance controller with an inner control loop that regulates the velocity of the object. The admittance is the outer loop that establishes the desired relation between the applied force to the object and its velocity. For the adaptation to a variety of loads, an online learning controller is implemented based on a neural network (NN) with backpropagation training. The overfitting of the NN is resolved with weight decay to decrease the oscillations around the equilibrium point. Alternatively, a gain scheduling PID controller is designed for the inner loop, which measures the object weight and tunes the gains with predefined rules. The performance of these two adaptation methods is demonstrated on an experimental setup and the results illustrate that better generalization can be achieved with the NN.

Nemec, Bojan; Vuga, Rok; Ude, Aleš

doi: 10.1080/01691864.2013.814211pmid: N/A

Abstract In this paper, we present a new approach to the problem of learning motor primitives, which combines ideas from statistical generalization and error learning. The learning procedure is formulated in two stages. The first stage is based on the generalization of previously trained movements associated with a specific task configuration, which results in a first approximation of a suitable control policy in a new situation. The second stage applies learning in the subspace defined by the previously acquired training data, which results in a learning problem in constrained domain. We show that reinforcement learning in constrained domain can be interpreted as an error-learning algorithm. Furthermore, we propose modifications to speed up the learning process. The proposed approach was tested both in simulation and experimentally on two challenging tasks: learning of matchbox flip-up and pouring.

Carbone, Giuseppe; D’Aliesio, Ettore; Borchert, Gunnar; Raatz, Annika

doi: 10.1080/01691864.2013.804800pmid: N/A

This paper describes a procedure for designing a novel mini-scaled parallel manipulator having three binary-actuated degrees of freedom, whose name is BAPAMAN2 (Binary Actuated PArallel MANipulator version 2). This is a low-cost, easy-operation manipulator that has been designed as an evolution of previous designs in the BAPAMAN series (BAPAMAN and BAPAMAN1). Main novel desired design features have been a smaller size and higher stiffness and repeatability performances. For this purpose, proper models have been developed and main design characteristics have been validated by means of numerical simulations. Experimental tests have been also carried out for validating the operation performances of a built prototype also in terms of stiffness and repeatability features.