A multi-strategy path planning method for mobile robots based on semantic maps

Abstract

The invention discloses a multi-strategy route planning method of a mobile robot based on a semantic map. Techniques such as a generalized Voronoi graph (GVG), conditional random field (CRF) inference and a morphological open and close operation are integrated in the method, noise elimination and semantic entitlement and the like on a map built by a common simultaneous localization and mapping (SLAM) algorithm are achieved, finally different route planning algorithms which are suitable for corresponding area features are applied to areas with different semantic meanings in the semantic map, and mixed use of the multi-strategy route planning algorithms is achieved. Based on semantic map planning navigation, not only is the superiority of a route used by a robot improved at the algorithm level, so that the capacity of the robot for selecting a route is inclined toward human intelligence more, but the multi-strategy route planning method lays a technical foundation for a good man-machine interaction of future robots.

Description

technical field [0001] The invention relates to a path planning method in the field of path planning for intelligent robots, in particular to a hybrid multi-strategy path planning method based on the establishment of a semantic map and applying different path planning algorithms in different areas of the map according to the semantic information. This enables the robot to autonomously select a path planning algorithm suitable for the environment in different motion environments more intelligently. Background technique [0002] At present, the research on robot systems capable of autonomous navigation and obstacle avoidance is extremely hot, and many products have been applied in practice. For example, a service robot in a restaurant has the ability to identify obstacles and plan paths; another example is a home sweeping robot, which also has the ability to plan its own movement path in an open space. In terms of public transportation, the application of drone express delive...

AI Technical Summary

Problems solved by technology

However, with the continuous expansion of robot application fields and more and more robot application scenarios, a single path planning strategy is difficult to meet people's increasing requirements for robots

The A* algorithm is a very good path planning algorithm. It performs well in environments with complex terrain and many obstacles such as rooms and offices, but it performs well in corridors and halls with fewer obstacles or long and thin roads. The performance in the environment is not satisfactory, because the path of A* is always close to the edge of the obstacle, but in the corridor, it is more inclined that the robot can walk in the middle of the road and pass through the slender corridor with higher safety

Images