Anh Son Phung

Input shaping and strain gauge feedback vibration control of an elastic robotic arm

Novel robotic applications, such as service robots, support the interaction between a human and a robot within the same workspace. These types of cooperative tasks require a safe robot operation in order to avoid physical harm to the human operator. In case of contact between the robot and human inherent safety is accomplished by means of lightweight flexible structures of reduced mass and stiffness that absorb contact forces. The flexible structure induces vibrations of the arm during motion, which complicates the precise kinematic control of the end effector pose. This contribution proposes input shaping in conjunction with strain gauge feedback control to suppress and compensate arm vibrations induced by robot motion. The control schemes are experimentally verified and the results demonstrate an efficient damping of link vibrations.

Vibration control of a multi-flexible-link robot arm under gravity

This paper proposes a novel independent joint control concept on a 3 DOF flexible link robot subject to deflections caused by gravity. The scheme dampens induced link oscillations in the presence of configuration dependant plant frequency and damping variations by integrating link strain feedback and impulse based input shaping. The approach is robust and the control does not depend on a dynamic model at runtime. The damping efficiency is evaluated experimentally in terms of strain measurement as well as end-effector position across the entire workspace.

Data based kinematic model of a multi-flexible-link robot arm for varying payloads

Reducing weight and inertias of conventional robot arms with an elastic structure allows safer interactive cooperation between humans and robots. While the end effector pose of a rigid robot is determined by the forward kinematic chain, the pose of elastic arms results from a superposition of the rigid kinematics and the pose dependent deflection caused by gravity. This property complicates the computation of forward and inverse kinematics in particular in case of dynamic loads. This paper presents a machine learning approach to extract various nonlinear regression models of the forward and inverse kinematics of a three degrees of freedom (DOF) flexible-link robot arm with dynamic loads from experimental data. The forward model predicts the target pose, given the joint angles and the strain signals while the inverse kinematic model predicts the joint angles required to assume a target pose. The transformation of the original features onto suitable nonlinear features substantially improves the generalisation ability of the both forward and inverse kinematic model. The closed loop inverse kinematic controller archieves a pose accuracy of 3 mm and the results show that the learned model can solve the inverse kinematics problem of flexible robot arms with sufficient accuracy even with unknown payloads.

Get Out of the Way - Obstacle Avoidance and Learning by Demonstration for Manipulation

Abstract Humans acquire manipulation skills by trial and error within a few trials, whereas programming a robot to perform the same task requires robotic expertise and effort. This paper presents a robot which learns a movement from demonstrations with the ability to generalize the movement to new goal poses and avoid the collision with obstacles in the workspace. The general movement is represented by dynamic movement primitives (DMP) augmented by potential fields in order to modulate the motion in the presence of obstacles. The approach is validated in experiments with a robotic arm in which dynamic obstacles partially blocking the movement are detected by a Photonic-Mixer-Devices (PMD) camera.

Tool centered learning from demonstration for robotic arms with visual feedback

Visual feedback allows controlling the relative pose between the camera and an object of unknown pose. This paper presents an approach to reproduce and generalize a movement from demonstrations based on visual feedback from an eye-in-hand camera mounted to the end effector. The main advantage of the approach is that no robot is required during the demonstration phase which is tool rather than robot centered. The teacher merely demonstrates the movement either with the camera alone or the camera attached to the tool that is required for the task. This method simplifies the demonstration in comparison to kinesthetic teaching that requires a compliant arm for motion recording. The accuracy and precision of the demonstration is enhanced as the teacher does not have to overcome the inertial and frictional forces of the arm while handling the end-effector in conjunction with the tool. The movement pattern in terms of translation and rotation of the camera with respect to the goal pose is described by dynamic movement primitives (DMP). Learning from demonstration is applied to straight point-to-point and more complex trajectories. The reproduced trajectories reveal a high precision independent from the object pose and the approach is confirmed by an experiment with a five degree-of-freedom (DOF) robot.

Predictive Delay Compensation for Camera Based Oscillation Damping of a Multi Link Flexible Robot

For oscillation damping of a multi link flexible arm under gravity this paper exploits the image data already acquired by an eye-in-hand camera used for visual servoing. It replaces commonly applied distributed strain measurements in a model free oscillation damping control concept. Based on simulations and experiments the paper compares three predictive signal processing approaches to compensate for the sensor inherent delay. Damping results for oscillations induced by joint motions as well as sudden load changes are presented in three different unstructured scenes.

Scene adaptive RGB-D based oscillation sensing for a multi flexible link robot arm in unstructured dynamic environments

The paper experimentally compares six visual oscillation sensing approaches for a three degrees of freedom flexible link robot arm with an eye-in-hand RGB-D camera. The comparison includes five representative scenarios. Based upon the results the authors propose a novel scene adaptive camera motion reconstruction scheme. The scheme adaptively selects the best approach according to the actual scene texture and depth profile. Experiments in indoor scenarios with sparse texture, poor depth profiles as well as dynamic scene contents approve the obtained signal quality to be well suited for visual vibration damping of flexible link robot arms in a great variety of frequently observed scenarios.

Learning to Catch Moving Objects with Reduced Impulse Exchange

Abstract The paper presents a learning from demonstration approach to the catching of moving objects with a robot manipulator. The work explicitly reduces the impulse exchange by minimizing the relative velocity during the contact phase by learning relative relation between the object and the catcher instead of learning separate forward model of the object and trajectory generator for the robot. This contributes to the damage prevention on both, the object and the robot. The demonstrated catching movements are modelled by a Gaussian mixture model (GMM), which describes the probability distribution over the demonstrated data set. Gaussian mixture regression (GMR) is employed for motion prediction. A timing controller is designed to trade-off the catch location and the catching time. The learning scheme comprises the relative position and velocity profile between the object and the catcher. The approach allows to generate the movement of the catcher with guaranteed position and velocity convergence to the reference inferred from the position and velocity of the object. Experimental results with a five degree of freedom arm and a Photonic-Mixing-Device (PMD) camera for object detection validate the approach.

Bildbasierte Regelung eines gliedelastischen Roboterarms mit einem RGB-D-Sensor

Zusammenfassung Der Beitrag befasst sich mit der bildbasierten Regelung eines mehrgliedrigen gliedelastischen Roboterarms unter Gravitationseinfluss anhand von RGB-D-Sensordaten. Der Fokus liegt auf der Mehrfachnutzung der in den zur Positionsregelung aufgenommenen Bildfolgen enthaltenen Bewegungsinformation auch zur Dämpfung der unerwünschten Armschwingungen. Es werden vier Ansätze zur Rekonstruktion der Schwingungsbewegung unter Verwendung der verfügbaren Tiefeninformation gegenübergestellt. Drei Ansätze zur Phasenkorrektur des bildbasiert am Endeffektor gewonnenen Schwingungssignals werden diskutiert. Die Validierung der bildbasierten Schwingungsdämpfung erfolgt im Experiment. Abstract The paper is focused on a vision based approach for the control of a multi-flexible-link robot arm under gravity using RGB-D-sensor data. Therefore the image sequences already acquired for conventional visual position control are exploited to extract a feedback signal for the damping of unwanted structural vibrations. The contribution compares four approaches for the visual vibration sensing, which benefit from the available depth information. Three concepts for the compensation of the phase lag inherent to the visual vibration signal are discussed. Finally the vision based vibration damping is validated experimentally.