Method, device and exploration method for intelligent agent to actively construct environmental scene map

Abstract

Provides a method for an agent to actively construct an environmental scene map based on visual information, an environment exploration method, and an intelligent device. The method includes: collecting the environmental scene image and the corresponding environmental scene map data set required for training the model; collecting the agent exploring the environment required for the training model path; the active exploration model is trained by using the environmental scene image and the corresponding environmental scene map data set and the collected agent to explore the environmental path; the action is generated based on the trained active exploration model, and the agent uses the generated action to explore the environment. , to obtain 3D semantic point cloud data, and then use the 3D semantic point cloud data to construct an environmental scene map. The invention can overcome the limitation that the traditional computer vision tasks can only passively perceive the environment in the past, and utilize the active exploration characteristics of the intelligent body to combine the perception ability and the movement ability to realize active perception, actively explore the environment, and actively construct the scene of the environment Atlas, applied to a variety of vision tasks.

Description

technical field [0001] The present invention generally relates to computer vision technology, and more specifically relates to a method, an intelligent device and an exploration method for actively constructing an environmental scene map by an intelligent body imitating a creature and a human. Background technique [0002] With the continuous development and wide application of machine learning technology, the field of computer vision has developed rapidly in recent years, and has achieved many remarkable results, including object detection, object recognition, and image semantic segmentation. However, these traditional 2D computer vision tasks can only passively perceive the environment and focus on analyzing the static environment, which ignores a very important feature of an agent (including creatures and humans): active exploration, that is, continuous Actively explore the environment to construct an environmental scene map and guide actions. Active exploration can prom...

AI Technical Summary

Problems solved by technology

However, these traditional 2D computer vision tasks can only passively perceive the environment and focus on analyzing the static environment, which ignores a very important feature of an agent (including creatures and humans): active exploration, that is, continuous Actively explore the environment to build an environmental scene map and guide actions

Images