Optical Wheel Tracking for Vehicle Motion Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomous and semi-autonomous vehicles face challenges in accurately tracking and predicting the motion of slow-moving vehicles, especially when they are oriented perpendicular to the host vehicle, as existing sensors like radar struggle to determine velocity under these conditions.

Innovation Solution

The method involves using camera image processing to identify and track the wheels of observed vehicles, determining their rotation rate and diameter, and calculating the vehicle's speed and direction of travel by analyzing the angular orientation of the wheels, which provides more accurate motion estimates and enhances navigation and collision avoidance systems.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If radar is used to detect velocity of observed vehicles, then detection range is improved, but measurement precision deteriorates when vehicles are oriented perpendicular to the host vehicle or moving slowly

Engineering Contradiction:
Improvedetection rangeVSAvoidvelocity measurement precision
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent uses the wheels of the observed vehicle as an intermediary object to infer vehicle motion. Instead of directly measuring vehicle velocity with radar, the system tracks wheel rotation through optical imaging, using the wheels as a mediator to obtain accurate speed and direction information even when the vehicle is perpendicular to the host vehicle or moving slowly

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces radar-based velocity detection with an optical imaging system that tracks wheel rotation. This substitution uses image processing and optical flow analysis to measure wheel rotation rates, converting the detection mechanism from electromagnetic wave-based radar to vision-based wheel tracking, which provides superior precision for perpendicular and slow-moving vehicles

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If optical tracking of wheel rotation is implemented, then speed and direction estimation accuracy is improved, but device complexity increases

Engineering Contradiction:
Improvespeed and direction estimation accuracyVSAvoidimage processing system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the observed vehicle into its wheel components for tracking purposes. By focusing specifically on the wheels rather than the entire vehicle, the system simplifies the tracking task while maintaining accuracy. The wheel features are identified and tracked independently, allowing for precise rotation measurement without needing to track the complex motion of the entire vehicle body

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent employs a multi-functional image processing system that simultaneously performs wheel detection, rotation rate calculation, vehicle speed estimation, and direction determination. This universal approach uses the same optical tracking infrastructure to achieve multiple measurement goals, reducing the need for separate specialized systems and thereby limiting the increase in device complexity

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

Data Source

PatentUS10706563B2State and position prediction of observed vehicles using optical tracking of wheel rotation
Publication Date: 2020.07.07 QUALCOMM INC
  • US10706563B2 patent drawing
  • US10706563B2 patent drawing
  • US10706563B2 patent drawing

AI summary

Various embodiments may include methods of using image data to estimate motion of a vehicle observed within camera images, such as images captured by a vehicle navigation system of a host vehicle. Various embodiments may include a camera capturing a sequence of images including the observed vehicle, and a processor performing image processing to identify a wheel of the observed vehicle, and determining a rate of rotation of the wheel based on changes in orientation of the wheel between at least two images within the sequence of images. The processor may further determine a speed of the observed vehicle based on the wheel's rate of rotation and diameter. The processor may further determine a direction of travel and/or turning rate of the observed vehicle by determining relative angles of wheels of the observed vehicle in at least one image.