Floor Detection Using Height Map Surface Normals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robots using LiDAR sensors face challenges in accurately detecting navigable floor space due to noise, calibration errors, and reflective surfaces, which can lead to erroneous perception of floor height and objects, causing navigation issues.
Innovation Solution
A robotic system that utilizes a controller to process depth measurements by projecting points onto a height map, calculating surface normal unit vectors, and determining floor pixels within a threshold deviation from a reference vector, effectively filtering out noisy data and improving floor detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional LiDAR sensors are used for floor detection, then the robot can obtain depth measurements, but the measurements are noisy and inaccurate due to noise, calibration errors, and reflective surfaces
Solution Approach 1:
The patent introduces a height map as an intermediary data structure between raw LiDAR depth measurements and floor detection. The height map projects 3D points onto a 2D grid with height values, serving as a mediator that organizes and smooths noisy measurements. This intermediary representation allows for more reliable floor detection by aggregating multiple measurements and reducing the impact of individual noisy points.
Solution Approach 2:
The patent transforms the problem from 3D space to 2D space by projecting depth measurements onto a height map. This dimensionality reduction simplifies the detection process and allows for more robust floor identification by working with height values in a 2D grid rather than processing raw 3D point clouds directly.
2Measurement precision
If the robot processes all depth measurements to detect floor, then comprehensive coverage is achieved, but computational complexity increases
Solution Approach 1:
The patent segments the processing task by dividing the environment into discrete height map cells. Each cell independently stores height information and can be processed separately. This segmentation allows for parallel processing and reduces the computational burden compared to analyzing all depth measurements globally, while still achieving comprehensive floor detection coverage.
3Measurement precision
If the robot uses surface normal vectors for floor detection, then accurate floor identification is achieved, but the calculation complexity increases
Solution Approach 1:
The patent changes the parameter representation from raw depth values to surface normal vectors. By computing normals from height map gradients, the system transforms the detection problem into comparing vector orientations against a reference vertical normal. This parameter transformation simplifies the decision logic while maintaining high accuracy in distinguishing floor from non-floor surfaces.
Data Source
AI summary
Systems and methods for detecting floor from noisy depth measurements for robots are disclosed herein. According to at least one non-limiting exemplary embodiment, a height map may be produced based on one or more depth measurements from a sensor of a robot. The height map may be utilized to determine surface normal vectors which may be further utilized by the robot to determine if regions of the height map are floor.