Robot Self-Localization Using Object Recognition and Environmental Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-localization methods for mobile robots using vision sensors are limited by the need for multiple sensors, inability to handle environmental changes, and inaccurate position estimation due to insufficient three-dimensional information.
Innovation Solution
A method employing a stereo video device with a camera unit, a database of topological maps, and particle filtering to recognize individual objects and estimate the robot's position based on three-dimensional information from the environment, allowing for accurate global position estimation even in dynamic environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a vision sensor is attached to the environment instead of the robot to recognize objects, then object recognition is achieved, but multiple vision sensors are required and the system becomes more complex
Solution Approach 1:
Instead of attaching vision sensors to the environment to recognize objects (prior art approach), this patent inverts the approach by attaching the vision sensor to the robot and using it to recognize objects in the environment. The robot-mounted vision sensor captures images, and the system performs object recognition by analyzing the captured images and matching them with stored object information in the database, thereby eliminating the need for multiple environment-mounted sensors.
Solution Approach 2:
The robot equips itself with a vision sensor and the capability to perform object recognition autonomously. The robot's own vision sensor captures environmental information, and the robot's processing system independently performs object recognition, position estimation, and navigation without requiring external sensors or infrastructure, making the system self-sufficient.
2Device complexity
If self-localization is performed based only on three-dimensional information of recognized objects, then position calculation is simplified, but position estimation becomes inaccurate when three-dimensional information is insufficient
Solution Approach 1:
This patent merges multiple sources of information for position estimation: it combines three-dimensional information of recognized objects with two-dimensional image information from the vision sensor. The system integrates object recognition results, environmental map data, and visual features to perform comprehensive position estimation, thereby improving accuracy without significantly increasing calculation complexity.
Solution Approach 2:
The patent transitions from relying solely on three-dimensional object information to incorporating two-dimensional image plane information as an additional dimension. By utilizing both 3D object data and 2D image features (such as object positions, sizes, and orientations in the image plane), the system achieves more robust and accurate position estimation, especially when 3D information alone is insufficient.
3Speed
If object recognition is performed while the robot is traveling, then continuous position tracking is achieved, but position estimation accuracy decreases due to motion blur and dynamic conditions
Solution Approach 1:
The patent implements periodic action by performing object recognition and position estimation at specific intervals or at predetermined positions along the robot's travel path rather than continuously during motion. The robot travels at normal speed and performs recognition operations periodically when conditions are favorable (e.g., when approaching known landmarks or at designated checkpoints), thereby balancing travel efficiency with accurate position estimation.
4Measurement precision
If artificial tags are attached to objects on the robot's traveling path to enable recognition, then object position detection is achieved, but the system becomes inconvenient and requires environmental modification
Solution Approach 1:
This patent extracts the artificial tag requirement from the system by using natural object features for recognition instead of requiring attached tags. The vision sensor captures images of objects with their natural appearances, and the system performs recognition by matching these images with stored reference information in the database, thereby eliminating the need for environmental modification through tag attachment while maintaining recognition accuracy.
Data Source
AI summary
A method for self-localization of a robot, the robot including a camera unit, a database storing a map around a robot traveling path, and a position arithmetic unit estimating the position of the robot, includes: acquiring an image around the robot, in the camera unit. Further, the method includes recognizing, in the position arithmetic unit, an individual object in the image acquired by the camera unit, to generate position values on a camera coordinate system of local feature points of the individual objects and local feature points of a surrounding environment including the individual objects; and estimating, in the position arithmetic unit, the position of the robot on the basis of the map and the position values on the camera coordinate system of local feature points of the individual objects and local feature points of a surrounding environment including the individual objects.