Pixelwise Depth Map Filtering for Robot Sensor Confidence

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Robotic systems face challenges in generating accurate and complete depth maps due to varying accuracy levels and resolutions from different sensors, leading to incomplete or inaccurate depth information, which affects navigation and interaction with the environment.

Innovation Solution

A pixelwise filterable depth map is generated by associating each pixel depth with a confidence level based on the sensor used, and corroborating depth information from multiple sensors, allowing for operation-specific depth maps by comparing pixelwise confidence levels with thresholds.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If depth information from multiple sensors with varying accuracy levels is combined, then the completeness of depth map coverage is improved, but the accuracy and reliability of depth information deteriorates due to inconsistent resolution and accuracy levels

Engineering Contradiction:
Improvedepth map completenessVSAvoiddepth information accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The depth map is segmented into multiple regions, each corresponding to a specific sensor's field of view and accuracy characteristics. Each region is labeled with metadata indicating the sensor source and confidence level, allowing the system to selectively use depth information from different sensors based on spatial location and operational requirements.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different regions of the depth map are assigned different quality attributes based on the sensor providing the data. High-accuracy sensors provide depth information for critical regions requiring precision, while lower-accuracy sensors fill in peripheral or less critical areas, optimizing the overall depth map quality for each specific region.

Inventive Principle:
Principle #3Local quality

2Device complexity

If a single depth map is generated for all robot operations, then the system complexity is reduced, but the adaptability to different robot operations deteriorates as different operations require different confidence levels

Engineering Contradiction:
Improvedepth map system complexityVSAvoidoperation-specific depth map adaptability
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The system generates multiple depth maps dynamically based on the current robot operation. Each depth map is filtered to include only pixel depths with confidence levels appropriate for the specific operation, allowing the depth map content and confidence thresholds to adapt in real-time to different operational requirements without requiring separate hardware systems.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

A single multi-sensor system and depth map generation architecture serves multiple robot operations with different confidence requirements. The same sensor array and processing pipeline produce operation-specific depth maps by applying different filtering criteria, eliminating the need for separate depth mapping systems for each operation type.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If all pixel depths from sensor data are used for robot navigation, then the processing speed is improved, but the reliability of navigation decisions deteriorates due to inclusion of inaccurate depth information

Engineering Contradiction:
Improvenavigation processing speedVSAvoidnavigation decision reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

Instead of processing all pixel depths uniformly, the system selectively processes only those pixel depths that meet the confidence threshold required for the current operation. This partial processing approach filters out low-confidence depth information that would not contribute meaningfully to reliable navigation decisions, reducing computational waste while maintaining processing efficiency for critical data.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3842888B1Pixelwise filterable depth maps for robots
Publication Date: 2026.04.15 GDM HOLDING LLC
  • EP3842888B1 patent drawingFigure 1
  • EP3842888B1 patent drawingFigure 2
  • EP3842888B1 patent drawingFigure 3

AI summary

A method includes receiving sensor data from a plurality of robot sensors on a robot. The method includes generating a depth map that includes a plurality of pixel depths. The method includes determining, for each respective pixel depth, based on the at least one robot sensor associated with the respective pixel depth, a pixelwise confidence level indicative of a likelihood that the respective pixel depth accurately represents a distance between the robot and a feature of the environment. The method includes generating a pixelwise filterable depth map for a control system of the robot. The pixelwise filterable depth map is filterable to produce a robot operation specific depth map. The robot operation specific depth map is determined based on a comparison of each respective pixelwise confidence level with a confidence threshold corresponding to at least one operation of the robot controlled by the control system of the robot.