Multi-layer Semantic Map Based Navigation for Mobile Robots in Agriculture Environment

Autonomous navigation has significant applications in agricultural environments. Based on the characteristics of agricultural environments, navigation is required to have a high level of intelligence, which necessitates making intelligent anthropomorphic decisions by comprehensively considering the actual environment. Traditional navigation methods often rely on the use of radar and cameras, which can lead to issues with adaptability and accuracy in practical applications. Agricultural environments often have many dynamic factors and environmental noise, and because they cannot distinguish between pedestrians and other dynamic obstacles, there may be potential resource wastage in path planning. This paper proposes an integrated semantic information layered cost map navigation system, which integrates different semantic information for different environments and provides it for use by the layered cost map. This enables it to make anthropomorphic decisions by comprehensively considering complex and variable environmental factors and the system's own characteristics. The use of layered cost maps is also conducive to the design of multi-agent collaborative systems.