Scholarlink

<p>The co-optimization of energy consumption and trajectory smoothness is a pivotal challenge for automated guided vehicle (AGV) path planning in industrial logistics. This paper introduces a novel deep reinforcement learning framework specifically designed to address this dual-objective problem. Our core contribution is an energy-conscious optimization mechanism that integrates a composite reward function with specialized constraints to simultaneously minimize energy use and ensure motion smoothness. The framework further enhances learning efficacy through an N-step bootstrap strategy and a gated recurrent network, enabling robust performance in dynamic environments. Comprehensive simulations in both static and dynamic industrial settings demonstrate the framework’s effectiveness. It achieves energy reductions of up to 56.03 and 17.81%, alongside smoothness improvements of up to 69.23 and 71.51%, respectively, compared to conventional baselines. These results validate the framework’s capability for tight dual-objective optimization within a unified learning agent, underscoring its practical potential for advancing sustainable and efficient industrial logistics.</p>

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

Energy consumption and trajectory smoothness co-optimization for automated guided vehicle path planning in industrial logistics

  • 期刊论文

  • 发表日期：

    2026年2月28日

查看全文
  • Haitao Yu,
  • Lidong Liang,
  • Yongjun Zhu,
  • Wenqiang Diao,
  • Gongwen Li

摘要

The co-optimization of energy consumption and trajectory smoothness is a pivotal challenge for automated guided vehicle (AGV) path planning in industrial logistics. This paper introduces a novel deep reinforcement learning framework specifically designed to address this dual-objective problem. Our core contribution is an energy-conscious optimization mechanism that integrates a composite reward function with specialized constraints to simultaneously minimize energy use and ensure motion smoothness. The framework further enhances learning efficacy through an N-step bootstrap strategy and a gated recurrent network, enabling robust performance in dynamic environments. Comprehensive simulations in both static and dynamic industrial settings demonstrate the framework’s effectiveness. It achieves energy reductions of up to 56.03 and 17.81%, alongside smoothness improvements of up to 69.23 and 71.51%, respectively, compared to conventional baselines. These results validate the framework’s capability for tight dual-objective optimization within a unified learning agent, underscoring its practical potential for advancing sustainable and efficient industrial logistics.