Monochromatic Camera Vehicle Pose Estimation Using Event-Based Optical Flow

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional vehicle pose estimation methods in enclosed spaces, such as parking garages, are costly and inefficient due to the need for multiple RGB cameras and sensors, and existing monochromatic camera-based systems fail to accurately detect and localize vehicles at a distance, especially when GPS signals are unavailable.

Innovation Solution

An event-based vehicle pose estimation system using a monochromatic camera that generates an optical flow map and pixel-level event mask, coupled with an unsupervised optical flow prediction network and Roll-Pitch-Yaw prediction network, to accurately detect and localize vehicles by converting monochromatic images into RGB patches and estimating vehicle pose only when a vehicle is detected, thereby reducing power consumption and costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple RGB cameras and sensors are used for vehicle pose estimation, then measurement precision is improved, but device complexity and cost increase

Engineering Contradiction:

Improvevehicle pose estimation accuracyVSAvoidnumber of cameras and sensors

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple functions (depth estimation, motion detection, pose estimation) into a single monochromatic camera system with event-based processing, eliminating the need for multiple separate RGB cameras and sensors while maintaining measurement precision through computational methods

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent replaces the mechanical/optical approach of using multiple physical cameras and sensors with a computational approach using a single monochromatic camera combined with event-based processing and neural network algorithms to achieve the same pose estimation functionality

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If continuous vehicle detection is performed, then measurement precision is maintained, but energy consumption increases

Engineering Contradiction:

Improvevehicle detection accuracyVSAvoidcamera system power consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent implements event-based triggering where the monochromatic camera only processes and captures images when motion events are detected, replacing continuous detection with event-driven periodic action, thereby maintaining detection accuracy while significantly reducing energy consumption during low-traffic periods

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The patent dynamically adjusts the detection and processing frequency based on real-time conditions, using event-based triggering to activate full processing only when vehicles are present, and entering low-power mode during idle periods, optimizing the balance between measurement precision and energy consumption

Inventive Principle:

Principle #15Dynamics

3Device complexity

If monochromatic cameras are used to reduce cost, then device complexity is reduced, but measurement precision deteriorates for distant vehicles

Engineering Contradiction:

Improvecamera system simplicityVSAvoiddistant vehicle localization accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces event-based processing and optical flow analysis as intermediary computational layers that enhance the information extracted from monochromatic images, allowing the simple hardware to achieve high measurement precision through sophisticated intermediate processing steps

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent changes the processing parameters and computational approaches (using event-based triggering, optical flow maps, and neural network-based pose estimation) to compensate for the limitations of monochromatic imaging, enabling accurate distant vehicle detection despite the simpler hardware

Inventive Principle:

Principle #35Parameter changes

4Measurement precision

If GPS signals are used for vehicle localization, then measurement precision is improved, but reliability deteriorates in enclosed spaces

Engineering Contradiction:

Improvevehicle localization accuracyVSAvoidsignal reception in enclosed spaces

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent enables the monochromatic camera system to independently perform pose estimation and localization without relying on external GPS infrastructure, using event-based processing and computer vision algorithms to provide self-contained localization that works reliably in GPS-denied enclosed environments

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS11482007B2Event-based vehicle pose estimation using monochromatic imaging

Publication Date: 2022.10.25 FORD GLOBAL TECH LLC

AI summary

A computer-implemented method for estimating a vehicle pose for a moving vehicle is described includes obtaining, via a processor disposed in communication with a monochromatic camera, a monochromatic image of an operating environment, and detecting in the monochromatic image an event patch showing a plurality of pixels associated with the moving vehicle. The method further includes generating an optical flow map using an unsupervised optical flow prediction network to predict an optical flow for each pixel in the monochromatic image. The optical flow map includes a Red-Green-Blue (RGB) patch having color information associated with a velocity for the moving vehicle. The system generates a pixel-level event mask that includes the RGB patch, and estimates the vehicle pose for the moving vehicle.