Depth Sensor Object Dimensioning With Adaptive Ray Error Bounds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sample consensus model fitting methods for determining object models do not adequately account for the inaccuracies of sensors, particularly near the edges of their field of view, leading to potential misclassification of data points as inliers or outliers.
Innovation Solution
A method and device that control a depth sensor to determine depth data, select an object model, define rays from the sensor to data points, and determine errors based on the distance to the object model's intersection, with stricter error thresholds near the sensor's field of view edges to reflect decreasing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a fixed error bound is used to classify data points as inliers or outliers, then the model fitting process is simple and fast, but the classification accuracy deteriorates near the edges of the field of view due to sensor inaccuracies
Solution Approach 1:
The patent applies local quality by making the error bound adaptive rather than fixed. The error bound varies locally across the field of view, being stricter near the center where sensor accuracy is higher and more lenient near the edges where sensor accuracy deteriorates. This resolves the contradiction by allowing fast processing (through automated adaptive bounding) while achieving accurate classification (through location-dependent error thresholds).
Solution Approach 2:
The patent implements dynamics by making the error bound dynamic and adaptive based on the data point's location in the field of view. Rather than using a static fixed bound, the system dynamically adjusts the error bound according to the radial distance from the field center, automatically adapting to local sensor accuracy characteristics. This enables both efficient processing and accurate classification.
2Measurement precision
If stricter error thresholds are applied near the sensor's field of view edges, then the classification accuracy improves, but the robustness of the model fitting deteriorates due to exclusion of valid data points
Solution Approach 1:
The patent resolves this contradiction by applying local quality through location-dependent error bounds. Data points near the center of the field of view are evaluated with stricter error thresholds, while data points near the edges are evaluated with more lenient thresholds. This ensures that each data point is judged by an appropriate standard for its location, maintaining both classification accuracy and model fitting robustness simultaneously.
Solution Approach 2:
The patent applies parameter changes by modifying the error bound parameter based on the spatial location of data points. The error bound parameter is changed as a function of radial distance from the field center, creating a gradient of acceptance criteria that matches the sensor's accuracy profile. This prevents both over-rejection of valid edge data and under-rejection of invalid center data.
Data Source
AI summary
A method comprising obtaining, from a sensor, depth data representing a target object; selecting a model to fit to the depth data; for each data point in the depth data: defining a ray from a location of the sensor to the data point; and determining an error based on a distance from the data point to the model along the ray; when the depth data does not meet a similarity threshold for the model based on the determined errors, selecting a new model and repeating the error determination for the depth data based on the new model; when the depth data meets the similarity threshold for the model, selecting the model as representing the target object; and outputting the selected model representing the target object.