Reproduce “CusADi2: Latency-Optimized Batch Computation of Robot Kinematics, Jacobians, and Hessians on the GPU” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

John Viljoen, Lasse Peters, Masayoshi Tomizuka, Negar Mehr

CusADi2 accelerates GPU evaluation of robot kinematics, Jacobians, and Hessians by scheduling independent operations within each computational graph in parallel, in addition to parallelizing across batch instances. It benchmarks 201 functions from 67 simulated robot models. At batch size 256, Hessian evaluation is a median 95× faster than CusADi, with gains strongest at intermediate batch sizes.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.