Scholarlink

<p>The proposed work presents a unique hybrid AI algorithm that gives a collision-free route for wheeled robots in an environment where more than one robot is present. This hybrid Neuro-GA-IWO algorithm combines Neural network, Genetic Algorithm, and Invasive Weed Optimization algorithms. Three individual algorithms are combined to obtain the optimum steering angle (OSA) as output. Six sensor data have been fed as input to the algorithm. The data collected from six sensors named FHD (Front Hindrance Distance), LHD (Left Hindrance Distance), RHD (Right-Front Hindrance Distance), LFHD (Left-Front Hindrance Distance), RFHD (Right-Front Hindrance Distance), and HA (Heading Angle). The hybrid algorithm was tested in both a simulation and a real environment. The output from the simulation deviated from real-time experimentation by less than 5% with the help of the developed hybrid algorithm. The developed algorithm can be implemented in both static and dynamic environments.</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

Neuro-GA-IWO hybrid AI-assisted trajectory planning approach for wheeled robots

  • 期刊论文

  • 发表日期：

    2025年8月20日

查看全文
  • Biswajit Sahoo,
  • Debasish Das,
  • Parnika Shrivastava,
  • Pinaki Das

摘要

The proposed work presents a unique hybrid AI algorithm that gives a collision-free route for wheeled robots in an environment where more than one robot is present. This hybrid Neuro-GA-IWO algorithm combines Neural network, Genetic Algorithm, and Invasive Weed Optimization algorithms. Three individual algorithms are combined to obtain the optimum steering angle (OSA) as output. Six sensor data have been fed as input to the algorithm. The data collected from six sensors named FHD (Front Hindrance Distance), LHD (Left Hindrance Distance), RHD (Right-Front Hindrance Distance), LFHD (Left-Front Hindrance Distance), RFHD (Right-Front Hindrance Distance), and HA (Heading Angle). The hybrid algorithm was tested in both a simulation and a real environment. The output from the simulation deviated from real-time experimentation by less than 5% with the help of the developed hybrid algorithm. The developed algorithm can be implemented in both static and dynamic environments.