Modeling of Computer Performance for Real-Time Parallel Grid-based Recursive Bayesian Estimation

Xianqiao Tong, Tomonari Furukawa, and Hugh F. Durrant-Whyte


recursive Bayesian estimation, real-time, no-Gaussian RBE, parallel


This paper presents the modeling of computer performance for the real-time parallel grid-based recursive Bayesian estimation (RBE). The proposed modeling formulates data transmission between the central processing unit (CPU) and the graphics processing unit (GPU) and floating point operations to be carried out in each CPU and GPU necessary for one iteration of RBE. Given the specifications of the computer hardware, the proposed modeling can thus estimate the total amount of time for RBE to perform RBE in a real-time environment. Prediction with separable convolution is further proposed to accelerate RBE for real-time operations. The performance of the proposed modeling was investigated, and parametric studies have first demonstrated its validity in various conditions by showing that the average error of estimation in computer performance stays below 6-7%. Utilizing prediction with separable convolution, the RBE has also been found to perform within 1 ms although the size of the problem was relatively large.

Important Links:

Go Back