Pisces: power-aware implementation of slam by customizing efficient sparse algebra
Published in 2020 57th ACM/IEEE Design Automation Conference (DAC), 2020
Citation: Bahar Asgari, Ramyad Hadidi, Nima Shoghi Ghaleshahi, Hyesoon Kim, 2020 57th ACM/IEEE Design Automation Conference (DAC), 1-6, 2020 https://ieeexplore.ieee.org/abstract/document/9218550/
This paper presents Pisces, a power-efficient approach to simultaneous localization and mapping (SLAM) for autonomous systems. By exploiting the sparsity of SLAM data, Pisces optimizes memory access patterns and enables pipelined and parallel processing, resulting in significant reductions in power consumption and execution time compared to state-of-the-art methods.