PulseAugur
LIVE 12:26:06
research · [2 sources] ·
0
research

New TSMC method optimizes trajectories and policies with differentiable dynamics

Researchers have introduced Tempered Sequential Monte Carlo (TSMC), a novel sampling-based framework for optimizing trajectories and policies within systems that have differentiable dynamics. This approach reframes controller design as an inference problem, aiming to minimize a KL-regularized expected trajectory cost. TSMC employs an annealing scheme to efficiently sample from complex target distributions by adaptively reweighting and resampling particles along a tempering path. The method has demonstrated broad applicability and superior performance compared to existing baselines in relevant benchmarks. AI

Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →

IMPACT Introduces a new optimization technique that could improve performance in robotics and control systems.

RANK_REASON This is a research paper describing a new method for trajectory and policy optimization.

Read on Hugging Face Daily Papers →

New TSMC method optimizes trajectories and policies with differentiable dynamics

COVERAGE [2]

  1. Hugging Face Daily Papers TIER_1 ·

    Tempered Sequential Monte Carlo for Trajectory and Policy Optimization with Differentiable Dynamics

    We propose a sampling-based framework for finite-horizon trajectory and policy optimization under differentiable dynamics by casting controller design as inference. Specifically, we minimize a KL-regularized expected trajectory cost, which yields an optimal "Boltzmann-tilted" dis…

  2. arXiv cs.LG TIER_1 · Heng Yang ·

    Tempered Sequential Monte Carlo for Trajectory and Policy Optimization with Differentiable Dynamics

    We propose a sampling-based framework for finite-horizon trajectory and policy optimization under differentiable dynamics by casting controller design as inference. Specifically, we minimize a KL-regularized expected trajectory cost, which yields an optimal "Boltzmann-tilted" dis…