Object Positioning with Camera-Rendered 3D Models in Weak-Signal Environments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vehicles and robots face challenges in accurate positioning due to inaccurate environment information obtained by sensors, leading to reduced positioning reliability, especially in conditions with weak signals like tunnels and bridge-tunnels.
Innovation Solution
An object positioning method that utilizes a camera to collect environment images, determines candidate position information and virtual camera parameters, renders a rendered image based on a three-dimensional model, and matches it with the environment image to determine accurate position information, leveraging a correspondence between virtual camera parameters and the real space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensors are used to obtain environment information for positioning, then positioning function can be implemented, but positioning accuracy deteriorates due to inaccurate sensor data
Solution Approach 1:
The patent creates a virtual copy of the real-world environment through a three-dimensional model. A virtual camera renders images based on candidate position information, producing a rendered image that corresponds to the actual environment. This rendered image is then matched with the real environment image captured by the camera, enabling accurate positioning without relying on inaccurate sensor data. The virtual copy serves as a reference to correct and verify sensor-based positioning results.
2Reliability
If multiple sensors are used to improve positioning reliability, then positioning function is enhanced, but device complexity increases
Solution Approach 1:
The patent introduces a three-dimensional model and virtual camera rendering system as an intermediary between the camera and the positioning result. Instead of directly processing and fusing data from multiple sensors, the system uses the pre-built three-dimensional model to generate expected environment images from candidate positions. This intermediary approach simplifies the positioning process by reducing it to an image matching problem, avoiding the complexity of multi-sensor data fusion while maintaining high reliability.
3Adaptability or versatility
If traditional sensor-based positioning is used, then positioning can be performed, but adaptability to severe environments deteriorates
Solution Approach 1:
The patent performs preliminary action by pre-building a three-dimensional model of the environment before positioning is needed. This model captures the geometric and visual characteristics of the environment in advance. When positioning is required, especially in severe conditions where sensor data may be insufficient, the system can immediately use this pre-prepared model to generate rendered images and perform matching, without needing to collect additional environment information from sensors.
Data Source
AI summary
An object positioning method includes obtaining an environment image collected by a target object through a camera of the target object and a camera parameter of collecting the environment image, determining candidate position information of the target object and a virtual camera parameter, rendering, based on a predetermined three-dimensional model, the candidate position information, and the virtual camera parameter, to generate a rendered image corresponding to the candidate position information, and determining position information of the target object based on the environment image, the rendered image, and the candidate position information. The virtual camera parameter and the camera parameter of the environmental image have a first correspondence. The three-dimensional model and a real space have a second correspondence.