Ceiling Contour Matching for Robotic Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic devices in warehouse environments face challenges in accurately localizing themselves within buildings due to the lack of reliable and unoccluded navigation features at ground level, necessitating the use of alternative stable and less occluded surfaces like ceilings for navigation.
Innovation Solution
Generating a depth map of static surfaces, such as ceilings, using sensors like stereo cameras or laser rangefinders, which allows robots to determine their position by aligning detected surface contours with corresponding contours in the map, utilizing the ceiling's contours for precise localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ground-level features are used for robotic navigation, then the navigation system can operate at robot level, but the localization accuracy deteriorates due to occlusions and lack of reliable features
Solution Approach 1:
Instead of using ground-level features for navigation as conventionally done, the patent inverts the approach by using ceiling surfaces as the reference for localization. The robot captures images of the ceiling, extracts contour features, and matches them against a pre-stored depth map of the ceiling contours to determine its position, thereby avoiding occlusions that affect ground-level features.
Solution Approach 2:
The patent transitions from two-dimensional ground plane navigation to three-dimensional ceiling surface navigation. By utilizing the vertical dimension and projecting ceiling contours onto the robot's coordinate system, the system creates a new navigation reference frame that is less susceptible to occlusions and provides more reliable localization features.
2Reliability
If ceiling surfaces are used for navigation, then localization accuracy improves due to stability and visibility, but the system complexity increases due to depth mapping requirements
Solution Approach 1:
The patent applies preliminary action by pre-generating and storing a depth map of the ceiling contours before the robot performs navigation. This depth map serves as a reference that the robot can match against real-time ceiling images, eliminating the need for the robot to perform complex real-time depth estimation and simplifying the onboard processing requirements.
Solution Approach 2:
The patent creates a digital copy of the ceiling surface in the form of a depth map with contour information. This copy is stored in the robot's memory and used for matching against actual ceiling images during navigation, allowing the robot to localize itself without requiring complex real-time depth sensing and processing capabilities.
Data Source
AI summary
An example method includes determining a depth map of at least one static surface of a building, where the depth map includes a plurality of surface contours. The method further includes receiving sensor data from one or more sensors on a robotic device that is located in the building. The method also includes determining a plurality of respective distances between the robotic device and a plurality of respective detected points on the at least one static surface of the building. The method additionally includes identifying at least one surface contour that includes the plurality of respective detected points. The method further includes determining a position of the robotic device in the building that aligns the at least one identified surface contour with at least one corresponding surface contour in the depth map.