Scholarlink

<p>Physical human–robot interaction (pHRI) requires controllers that ensure both safety and efficiency while adapting to dynamic human inputs. This paper evaluates a neuroadaptive admittance controller (NAC) for ambulation through pHRI, implemented on the Adaptive Robotic Nursing Assistant (ARNA), a mobile manipulator designed for physical assistance and task automation in healthcare settings. The proposed NAC is a neural network-based adaptive torque controller designed to facilitate physical interaction with the ARNA robot. This paper outlines the design and subsystems of the ARNA robot, focusing on its mobile base and handlebar components. Additionally, we present the admittance control strategy, which regulates the robot’s mechanical compliance. This strategy is formulated and experimentally tested in conjunction with the NAC. We conducted two sets of experiments: First, our NAC controller was compared to a classical proportional derivative (PD) velocity controller through human user experiments. Secondly, more experiments were conducted with a fine-tuned NAC and expert insights from nursing practitioners. The results demonstrate a further performance improvement. Our findings indicate that the NAC outperforms the PD controller in terms of accuracy, efficiency, user experience, and overall smoothness of the interaction.</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

Evaluation of a neuroadaptive admittance controller for ambulation through physical human–robot interaction

  • 期刊论文

  • 发表日期：

    2025年11月5日

查看全文
  • Payman Sharafianardakani,
  • Shamsudeen Abubakar,
  • Sumit K. Das,
  • Shikha Surupa,
  • Mohamed A. Hanafy,
  • Madan Rayguru,
  • Dan O. Popa

摘要

Physical human–robot interaction (pHRI) requires controllers that ensure both safety and efficiency while adapting to dynamic human inputs. This paper evaluates a neuroadaptive admittance controller (NAC) for ambulation through pHRI, implemented on the Adaptive Robotic Nursing Assistant (ARNA), a mobile manipulator designed for physical assistance and task automation in healthcare settings. The proposed NAC is a neural network-based adaptive torque controller designed to facilitate physical interaction with the ARNA robot. This paper outlines the design and subsystems of the ARNA robot, focusing on its mobile base and handlebar components. Additionally, we present the admittance control strategy, which regulates the robot’s mechanical compliance. This strategy is formulated and experimentally tested in conjunction with the NAC. We conducted two sets of experiments: First, our NAC controller was compared to a classical proportional derivative (PD) velocity controller through human user experiments. Secondly, more experiments were conducted with a fine-tuned NAC and expert insights from nursing practitioners. The results demonstrate a further performance improvement. Our findings indicate that the NAC outperforms the PD controller in terms of accuracy, efficiency, user experience, and overall smoothness of the interaction.