Large Language Modeling and Visual Perception Based Home Service Robot Research

Aiming at the problems of single dataset instruction and inaccurate intention recognition in the practical application of home service robots, a method based on yolov8 target detection and using LLM technology to realize the intelligence of the object from target detection to motion planning execution is proposed. Firstly, target detection using yolov8 network realizes visual perception of objects; secondly, the large language model is fine-tuned to adapt to a variety of specific tasks of home service robots; thirdly, motion planning research is carried out on the robotic arm to complete the sequence generation of the end-executed actions. Finally, a conclusion is made and an outlook is given.