Scholarlink

In the present work, the optimization problems arising during the generation of human-like movements for an anthropomorphic robot performing shelf replenishment tasks are evaluated in detail. Specifically, a set of movements that involve grasping five objects from a support table and transporting each object to a shelf is analyzed. These movements are formulated as nonlinear optimization problems, with various constraints related to the posture of the arm and the objects in the workspace. The impact of different model simplifications on the computational time necessary for the Interior Point OPTimizer (IPOPT) solver to obtain an optimal solution is explored. The optimization problems are modeled using the A Mathematical Programming Language (AMPL) modeling language and different levels of presolve are used. Our results demonstrate that increasing the presolve level enhances computational efficiency. However, excessively high presolve may negatively affect performance for some problems when compared to lower levels. This research highlights the importance of selecting an appropriate presolve level to optimize both problem formulation and solution time.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Study on the Effects of Model Simplifications for Generating Human-Like Robot Movements in Refill Tasks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Daniel Rodrigues,
  • Eliana Costa e Silva,
  • Gianpaolo Gulletta,
  • Pedro Ribeiro,
  • Inês Costa,
  • Wolfram Erlhagen,
  • Cátia Ferreira,
  • Luís Louro,
  • Estela Bicho

摘要

In the present work, the optimization problems arising during the generation of human-like movements for an anthropomorphic robot performing shelf replenishment tasks are evaluated in detail. Specifically, a set of movements that involve grasping five objects from a support table and transporting each object to a shelf is analyzed. These movements are formulated as nonlinear optimization problems, with various constraints related to the posture of the arm and the objects in the workspace. The impact of different model simplifications on the computational time necessary for the Interior Point OPTimizer (IPOPT) solver to obtain an optimal solution is explored. The optimization problems are modeled using the A Mathematical Programming Language (AMPL) modeling language and different levels of presolve are used. Our results demonstrate that increasing the presolve level enhances computational efficiency. However, excessively high presolve may negatively affect performance for some problems when compared to lower levels. This research highlights the importance of selecting an appropriate presolve level to optimize both problem formulation and solution time.