Ruediger Dillmann

Grasp planning: Find the contact points

Automatically grasp planning for dexterous hand with a large number of degrees of freedom (DOF) is still a challenging problem. Two groups of grasp quality measurements can be considered: one associated with the position of contact points and the other associated with the hand configuration. In this paper, we propose a novel approach to bridge the gap between the two grasp quality measurements by finding all possible contact points between a robotic hand and an object with given position and orientation. This way the contact points fulfill the second group of measurements. The first group of measurements can be used to find the best grasp, which could be executed on a real hand without unreachable or joint limit problems. We use a continuous collision detection algorithm and swept volume to efficiently gather all possible contact points through the entire configuration space of each finger. Results of finding contact points between the Schunk Anthromorph Hand (SAHand) with 13 DOFs and various objects in every- days environment are presented.

Range Image Registration Using an Octree based Matching Strategy

Autonomous world modeling is one of the major topics in current robot research. A basic concept hereby is the registration of consecutive range images. Consistent models can only be built with robust registration methods. Already one incorrect registered range image will affect the following registrations and lead to an inconsistent model. Therefore we developed a robust ICP registration method based on an octree matching strategy. This matching strategy could cope with large odometry errors and achieved the generation of consistent 3D models.

Social mechanisms of robot programming by demonstration

Keywords: Learning by imitation ; Robotics Reference LASA-BOOK-2007-002 URL: http://www.sciencedirect.com/science?_ob=PublicationURL&_tockey=%23TOC%235666%232006%23999459994%23621860%23FLA%23&_cdi=5666&_pubType=J&view=c&_auth=y&_acct=C000013218&_version=1&_urlVersion=0&_userid=164550&md5=3370f278f67f61eced6a66009be198b4 Record created on 2007-11-14, modified on 2017-05-10

Autonomous grasp and manipulation planning using a ToF camera

A time-of-flight camera can help a service robot to sense its 3D environment. In this paper, we introduce our methods for sensor calibration and 3D data segmentation to use it to automatically plan grasps and manipulation actions for a service robot. Impedance control is intensively used to further compensate the modeling error and to apply the computed forces. The methods are further demonstrated in three service robotic applications. Sensor-based motion planning allows the robot to move within dynamic and cluttered environment without collision. Unknown objects can be detected and grasped. In the autonomous ice cream serving scenario, the robot captures the surface of ice cream and plans a manipulation trajectory to scoop a portion of ice cream.

Graspability: A description of work surfaces for planning of robot manipulation sequences

For complex manipulation with multiple objects a service robot needs information about the structure of its environment including how and where it can manipulate in it. For this purpose, we introduce the Graspability. It is a measure describing the quality of a pose in Cartesian space for grasping or placing an object. The graspability considers kinematic reachability for a grasping robot and available grasps for the object. It is based on the assumption, that in manipulation tasks, objects tend to be located on a planar surfaces and have to be graspable from that plane. based on that assumption, we develop a discrete map of the environment which enables the use of the graspability in an autonomous planning system for complex manipulation tasks with multiple objects. Generated manipulation actions are evaluated on a real robot.

Experimental evaluation of the schunk 5-Finger gripping hand for grasping tasks

In order to perform useful tasks, a service robot needs to manipulate objects in its environment. In this paper, we propose a method for experimental evaluation of the suitability of a robotic hand for grasping tasks in service robotics. The method is applied to the Schunk 5-Finger Gripping Hand, which is a mechatronic gripper designed for service robots. During evaluation, it is shown, that it is able to grasp various common household objects and execute the grasps from the well known Cutkosky grasp taxonomy [1]. The result is, that it is a suitable hand for service robot tasks.

A ray-shooting based quality measurement for grasping and manipulation

If the manipulation task information is available at the time of planning, such as forces and torques that a grasp should apply onto the object, grasps more suitable for the task can be found. In this paper, we propose a ray-shooting based algorithm, which computes the quality of a grasp for performing given forces and torques. The method can deal with both force-closure grasps and non force-closure grasp. In case of multiple forces and torques, task wrench space is built. The quality is measured as the scale that the task wrench space fit into the grasp wrench space. Numerical experiments show the the feasibility of our method.

Efficient grasp planning with reachability analysis

Grasping can be seen as two steps: placing the hand at a grasping pose and closing the fingers. In this paper, we introduce an efficient algorithm for grasping pose generation. Depend on the hand kinematic, boxes of different sizes are sampled. The reachability for graping is represented by the information, from where the hand can grasp the box firmly. These boxes represent real objects, which at run-time will be decomposed into such boxes, so that the grasping poses for the real object can be generated. Concrete grasps at a grasping pose will be further checked for its grasp quality. Real experiments with two different robotic hands show the efficiency and feasibility of our method.

Exact spike timing computational model of convolutional associative memories

In this paper, we propose the implementation of an associative memory model using exact spike time signalling. We implement Circular Holographic Reduced Representations, an implementation of Vector Symbolic Architectures, which is a candidate for an associative memory framework described by holonomic brain theory. Using this architecture, we demonstrate the capacity of spiking neural networks to implement an associative memory with a high degree of recall accuracy. The recordings of spiking dynamics during various computational stages contain elements of the stochastic spike synchrony observed in cortical EEG recordings, as well as clearly separable computational phases suitable for functional analysis. We show that the framework is able to very accurately memorize and recall associated symbolic identifiers under ideal conditions (with no noise), maintaining very accurate and predictable spike timing. Furthermore, we test the framework in the noisy environment which is known to exist in cortical tissue. Besides reporting accuracy analysis for this case, we propose computational requirements (constraints) needed for the exact spike timing framework to work in such noisy conditions.

Monitoring of manipulation activities for a service robot using supervised learning

To be a good helper, grasping and manipulation are the most important abilities of a service robot. It should be able to adapt its manipulation actions to new tasks and environments. During the execution, it is important to rate the success of actions, so that the robot can plan and execute further actions to correct and recover from the failed actions. The successful execution of manipulation actions depends on various factors during the whole execution, such as the position of the robotic hand and forces exerted by the robot. The goal of the manipulation action monitoring is to estimate the success state from the huge amount of data collected during the execution. The main challenge to solve this problem is to identify the success or failure state from the the high dimensional data collection. We propose a method to classify ongoing activities using a set of support vector machines (SVM). After a supervised training process with manually labeled successful or failure results, our system can correctly estimate the resulting state of a manipulation activity. We present experiments on our bimanual manipulation demonstrator and evaluate the results.