Mobile robot environment map construction method and system and storage medium
A mobile robot, environmental map technology, applied in control/regulation systems, radio wave measurement systems, instruments, etc., can solve the problems of redundant coverage, slow environmental space, and no balance between benefits and costs, and achieves a reduction in redundant coverage. Effect
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Publication Date
- 2020-12-04
- Estimated Expiration
- Not applicable · inactive patent
Smart Images

Figure 1 
Figure 2 
Figure 3
Abstract
Description
technical field
[0001] The invention belongs to the technical field of mobile robot autonomous map construction, in particular to a mobile robot environment map construction method, system and storage medium. Background technique
[0002] In recent years, from home services to disaster relief and reconnaissance to alien exploration missions, the development of robots has greatly facilitated human life, industrial manufacturing, scientific research and military activities. The basic element for a mobile robot to successfully complete a specific task is accurate perception of the environment, which includes building a complete and accurate map. Traditional mapping research focuses on map representation, map fusion, and efficient map storage methods, but pays little attention to the autonomy of robotic mapping. Environmental maps are often collected and constructed by remote-controlled robot movement or by letting the robot move randomly in the environment. Some robots with a...
Examples
Embodiment Construction
[0034] The purpose of the present invention is to provide a method for a mobile robot to use deep reinforcement learning to build a map autonomously in an unknown environment, such as figure 2As shown, it obtains perception data from the environment through its own lidar sensor, and then constructs a two-dimensional grid map of the environment from the known sensor data, and uses a boundary-based method to detect the gap between free space and unexplored space. The boundary point, and then select an optimal boundary point from all the current boundary points based on the income and cost. The optimal boundary point is the target position of the robot's movement, and then use the deep reinforcement learning method to control the robot to realize the autonomous obstacle avoidance of the mobile robot The navigation moves to the boundary point, obtains new environmental information, and performs a new round of mapping. This process is repeated until there are no boundary points in...