Random Number Generators and Sphere Sampling for High-Level Synthesis

Randomized algorithms are often utilized for applications such as path planning in autonomous mobile systems. FPGAs are considered a solution candidate for the computation to cope with the heavy load caused by the huge number of iterations. This paper presents our designs of Random Number Generators (RNGs) for FPGAs, including Mersenne Twister, Combined Tausworthe, and Sphere Sampling. We utilize High-Level Synthesis (HLS) design tools for productivity and tune the code considering the resultant architecture for performance.