Virtual Fence Environment Perception Algorithm for Industrial Robots Based on 3D Vision

In order to solve the inconvenient problem of installing a physical fence for robots, especially the problem that the physical fence does not solve the problem of making robots actively perceive the environment, this paper proposes a multi-level virtual fence control system based on an RGBD camera. This strategy generates the spatial virtual fence point cloud in advance, and the virtual fence exists in the space of multiple levels. After the camera was installed above the robot and hand-eye calibration was carried out, the ambient point cloud captured by the RGBD camera was mapped to the coordinate system of the robot base. The ground is removed, the objects are extracted from the environmental point cloud, and the objects are segmented into clusters, and then the fence is divided into layers to construct KD-tree. In terms of object judgment, an improved KNN hierarchical search model with weight assignment of control points is proposed. Experiments show that the method is effective.