Distributed visual sensing network-based movable navigation system

Abstract

The invention discloses a distributed visual sensing network-based movable navigation system, which comprises a visual network system and a movable object. The visual network system is mainly formed by connecting a plurality of visual sensors through wireless or wired networks, and is used for providing an intelligent environment; and the movable object is provided with a control box capable of being communicated with the visual network system, and is controlled to move by the visual network system. The system realizes navigation of non-intelligent vehicles in an extensive and strange environment by building the intelligent environment, not only has high reliability, but also is economical and practical.

Description

technical field [0001] The invention belongs to the technical field of communication and control, and relates to a mobile navigation system, in particular to a mobile navigation system based on a distributed visual sensor network. Background technique [0002] Existing mobile robots and smart car navigation usually rely on an on-board central computer and vision system to imitate the human brain to achieve centralized environment detection and information processing, so that the robot can acquire autonomous movement and behavior capabilities. However, due to the dynamic change and unstructured environment, such research faces a bottleneck problem of implementation complexity. For example, in order to accomplish autonomous navigation, considerable effort is required in cognition and knowledge representation, requiring considerable computational effort, with visual road signs to answer questions such as "Where am I?" and "What should I do now? "Have I been to this place?" Bas...

AI Technical Summary

Problems solved by technology

[0003] 1. As far as image processing and scene understanding are concerned, the projection of the 3D world onto the 2D image plane through the pinhole projection model (used only by vertebrates) makes motion analysis itself pathological, such as rotation and movement in an impossible way. The way of disaggregation is mixed, which makes it difficult to achieve effective and reliable robot navigation

[0004] 2. As far as motion control is concerned, the arrangement of the head camera introduces more uncertainty due to the change of perspective caused by the movement of the camera

This makes the link between planning and control quite complex

[0007] 5. In terms of task formulation, discovering semantics from visual sensors is complex for real-time applications. To represent and interpret semantics, a task-level plan requires a lot of computation and storage space

For this reason, some navigation technologies that artificially change the environment to reduce the difficulty of centralized processing have been applied in practice, such as laser-guided and magnetic-track-guided automatic guided vehicles, but currently these solutions can only work in relatively fixed and simple environments, such as Parts handling between production lines in factories

Images