Scholarlink

This paper proposes an intelligent kinematic controller for autonomous ground vehicles (AGVs) using a Self-Organizing Generalized Regression Neural Network (SOGRNN). The SOGRNN controller updates its structure and parameters online in response to environmental dynamics, allowing it to adapt in real time through operations such as adding, pruning, and replacing nodes. The focus of this paper is on the self-organizing capability of the controller, which enables autonomous systems to navigate in outdoor terrains with unknown external disturbances. The SOGRNN is utilized to control the velocity of the vehicle, generating suitable control commands based on kinematic information at each way-point while the jackal AGV is in motion. It successfully produces appropriate linear and angular velocity control signals. To validate the performance of the controller, experiments are conducted in both simulated and real AGV environments. The implementation and demonstration take place in the Gazebo simulation environment and on real AGV platforms, namely the Pioneer3dx and Jackal, utilizing the Robot Operating System (ROS). The experiments are carried out in outdoor terrains to reflect real-world conditions. Finally, the performance of the kinematic controller in the autonomous path planning module is experimentally validated, encompassing both simulated and actual AGV scenarios. This serves as evidence of the effectiveness of the proposed controller in enabling autonomous navigation in outdoor environments with unknown disturbances.

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

SOGRNN Controller for Path Planning Module of WMR: Real-Time Adaptation and Navigation in Outdoor Environments

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sobers L. X. Francis,
  • Ahmad Jobran Al-Mahasneh,
  • Sreenatha G. Anavatti,
  • Matthew Garratt,
  • Osama Hassan

摘要

This paper proposes an intelligent kinematic controller for autonomous ground vehicles (AGVs) using a Self-Organizing Generalized Regression Neural Network (SOGRNN). The SOGRNN controller updates its structure and parameters online in response to environmental dynamics, allowing it to adapt in real time through operations such as adding, pruning, and replacing nodes. The focus of this paper is on the self-organizing capability of the controller, which enables autonomous systems to navigate in outdoor terrains with unknown external disturbances. The SOGRNN is utilized to control the velocity of the vehicle, generating suitable control commands based on kinematic information at each way-point while the jackal AGV is in motion. It successfully produces appropriate linear and angular velocity control signals. To validate the performance of the controller, experiments are conducted in both simulated and real AGV environments. The implementation and demonstration take place in the Gazebo simulation environment and on real AGV platforms, namely the Pioneer3dx and Jackal, utilizing the Robot Operating System (ROS). The experiments are carried out in outdoor terrains to reflect real-world conditions. Finally, the performance of the kinematic controller in the autonomous path planning module is experimentally validated, encompassing both simulated and actual AGV scenarios. This serves as evidence of the effectiveness of the proposed controller in enabling autonomous navigation in outdoor environments with unknown disturbances.