BORA: A Bag Optimizer for Robotic Analysis

Authors: Jian Zhang (ShanghaiTech University); Tao Xie (San Diego State University); Yuzhuo Jing, Yanjie Song, and Guanzhou Hu (ShanghaiTech University); Si Chen (West Chester University of Pennsylvania); and Shu Yin (ShanghaiTech University)

Abstract: We present BORA, a file system middleware that optimizes the acquisition of bags, which are the robotic operating system (ROS) formatted files. BORA sits between ROS and an existing file system to conduct semantic-aware data pre-processing. It categorizes bag data into groups with each having a distinct label. BORA predigests data index constructions and reduces file open time via a hash-based label management scheme. We implement a BORA prototype and then integrate it into a single-node server, a four-node PVFS storage cluster and a production cluster. Next, we evaluate the BORA prototype using four real-world ROS applications. Experimental results show that compared to a traditional bag management scheme, a BORA-assisted file system improves data acquisition performance by up to 11x. It also offers up to 10x data acquisition performance improvement and 3100x bags open improvement under a swarm robotics data analysis scenario where data is retrieved across multiple bags simultaneously.




Back to Technical Papers Archive Listing