With the emergence of cloud robotics, the cloud computing paradigm becomes increasingly attractive to robotics, where the cloud acts as the remote brain of low-cost robots, such as commodity drones. The idea is to offload heavy computations, like image processing, from the robot to the cloud; process it in short time (near real-time) and send back commands to the robot. This paper investigates the performance of a back-end cloud computing framework in deploying robotics-like applications (i.e. image analysis and processing) using low-cost Hadoop clusters. The design of a low-cost mini-data center built with readily available commodity 32-bit ARM boards, i.e. Raspberry Pi 2 Model B, is presented. Furthermore, the performance of RPi-based clusters is extensively tested with different types of data including text, text/image and image, and a comparative analysis against Hadoop cluster running on virtual machines is presented. The Hadoop Image Processing Interface (HIPI) Library was used and also configured to optimally utilize the Pi Cluster resources for improved performance. Results show that the RPi Hadoop cluster lags in performance when compared to Hadoop cluster running on virtual machines, the low cost and small form factor makes it ideal for remote Image analysis in surveillance / disaster recovery scenarios where UAVs can transmit image streams to the Cluster for remote processing. ; info:eu-repo/semantics/publishedVersion


    Access

    Download


    Export, share and cite



    Title :

    Performance of a Low Cost Hadoop Cluster for Image Analysis in Cloud Robotics Environment


    Contributors:

    Publication date :

    2016-01-01


    Remarks:

    doi:10.1016/j.procs.2016.04.013



    Type of media :

    Article (Journal)


    Type of material :

    Electronic Resource


    Language :

    English



    Classification :

    DDC:    629



    Implementation of Big Data in Cloud Computing with optimized Apache Hadoop

    Thind, Jaskaran Singh / Simon, Rajbala | IEEE | 2019





    Compression Accelerator for Hadoop Appliance

    Kim, S.D. / Lee, S.M. / Jang, J.H. et al. | British Library Conference Proceedings | 2014