No abstract

With the goal of advancing the state of automatic robotic grasping, we present a novel approach that combines machine learning techniques and rigorous validation on a physical robotic platform in order to develop an algorithm that predicts the quality of a robotic grasp before execution. After collecting a large grasp sample set (522 grasps), we first conduct a thorough statistical analysis of the ability of grasp metrics that are commonly used in the robotics literature to discriminate between good and bad grasps. We then apply Principal Component Analysis and Gaussian Process algorithms on the discriminative grasp metrics to build a classifier that predicts grasp quality. The key findings are as follows: (i) several of the grasp metrics in the literature are weak predictors of grasp quality when implemented on a physical robotic platform; (ii) the Gaussian Process-based classifier significantly improves grasp prediction techniques by providing an absolute grasp quality prediction score from combining multiple grasp metrics. Specifically, the GP classifier showed a 66% percent improvement in the True Positive classification rate at a low False Positive rate of 5% when compared with classification based on thresholding of individual grasp metrics.978-1-4799-6934-0/14/$31.00 ©2014 IEEE

No abstract

In this paper, we propose the addition of passive hydraulic mechanisms to simple parallel robotic grippers for improving disturbance rejection while maintaining the low cost of an industry standard gripper design. Each adaptive jaw on our gripper consists of three parallel hydraulic cylinders that are connected to a common local reservoir. The resultant passive hydraulic system is fully encased in the finger and moves independently of the actuator that closes the fingers. Such a design eliminates the need to engineer a complex cable or linkage system to allow for finger adaptability as many underactuated grippers do. Specifically, hydraulic cylinders need only be selected and connected together. As with other underactuated devices, the unconstrained freedoms of this design allow the gripper to adapt to unknown objects instead of creating a custom gripper shape for each new object the robot needs to grasp. In this paper, we analyze the ability of this gripper to maximize contact points over various sized objects and object placements while creating immobilizing form closure grasps. We than tested these improvements on a physical robot and found that grasp performance increased by up to 30% over a gripper lacking underactuation.

For the U.S. Department of Defense, high-performance power sources are a critical technology for current and future gun-fired munitions. Molten salt thermal batteries are the reserve power source of choice for many weapon systems due to their high power density, proven long shelf life, and capability to function across a wide range of operating environments. The incorporation of nanomaterials into thermal battery components offers the potential to produce batteries with performance improvements, such as higher voltages at increased current densities. In this study, nanoscale iron disulfide (FeS2) was synthesized using a fully scalable, high energy milling method, and characterized utilizing SEM, XRD, ICP-OES, BET surface area analysis, TGA-DSC, and other techniques. Nanostructured cathode mixtures containing the nanoscale FeS2 and LiCl-KCl eutectic salt electrolyte were pressed into electrode pellets. Their electrochemical performance was then characterized as compared to standard FeS2 cathode electrode pellets currently used in industry. Single cell thermal battery discharge testing, polarization type scans, and pulse load capability were used to compare cell performance for better understanding of best-use applications for these materials. Compared to control cells, the cells utilizing nanostructured cathodes provided higher power response, with respect to both voltage and runtime.