Computer Vision Pipeline for Autonomous Motion Resource Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomously motile devices face challenges in efficiently processing image data to determine object properties and perform actions based on these properties, particularly in optimizing the use of computer vision components for specific commands, leading to unnecessary resource consumption and inefficiency.

Innovation Solution

The device determines and configures specific computer vision components to process image data only in response to particular commands, using a pipeline approach where each component processes data based on the output of previous components, optimizing resource usage by only activating relevant components for the required tasks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all computer vision components are activated to process image data, then comprehensive object detection and tracking are achieved, but resource consumption increases and processing efficiency decreases

Engineering Contradiction:
Improveobject detection accuracyVSAvoidresource consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system dynamically configures computer vision components based on command types and environmental conditions. Different command types (e.g., object detection, tracking, classification) activate different subsets of vision components, allowing the system to adapt its processing power to the specific task at hand, thereby reducing unnecessary resource consumption while maintaining detection accuracy.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The computer vision system is divided into multiple independent components that can be selectively activated. Each component handles specific aspects of image processing (e.g., feature extraction, object detection, tracking), and only the relevant components are engaged based on the current command requirements, preventing full-system activation for every task.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If multiple computer vision components are configured to process image data, then comprehensive object analysis is achieved, but processing time increases

Engineering Contradiction:
Improveobject property determination accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system applies partial action by activating only the minimum necessary computer vision components required to fulfill the specific command. For example, simple object detection commands may only require basic feature extraction, while more complex tasks like detailed classification or tracking may engage additional components. This selective approach reduces processing time while maintaining sufficient accuracy for the given task.

Inventive Principle:
Principle #16Partial or excessive action

3Speed

If computer vision components are continuously activated, then real-time object tracking is maintained, but energy consumption increases

Engineering Contradiction:
Improveresponse speedVSAvoidenergy consumption
Core Design Contradiction:
SpeedVSLoss of energy

Solution Approach 1:

Instead of continuous activation, the system employs periodic or event-driven activation of computer vision components. Components are activated based on specific triggers such as command reception, detected object significance, or time intervals, allowing the system to maintain readiness for real-time responses while avoiding constant energy-consuming operation of all vision components.

Inventive Principle:
Principle #19Periodic action

Data Source

PatentUS11412133B1Autonomously motile device with computer vision
Publication Date: 2022.08.09 AMAZON TECH INC
  • US11412133B1 patent drawing
  • US11412133B1 patent drawing
  • US11412133B1 patent drawing

AI summary

A device capable of autonomous motion may process image data determined by one or more cameras to determine one or more properties of objects represented in the image data. The device may determine that two or more computer vision components correspond to a particular property. A first computer vision component may process the image data to determine first output data, and the second computer vision component may process the first output data to determine second output data corresponding to the property.