AMR Human Action Detection for Unintended Motion Safety

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In autonomous mobile robot (AMR) environments, unintended human actions can lead to catastrophic consequences due to misinterpretation, increasing the risk of collisions and damage, especially as AMRs become more autonomous and interact closely with humans and infrastructure.

Innovation Solution

A data-driven unintended human action detector using a deep neural network that incorporates current human state, predicted human actions, and AMR sensor information, combined with a rule-based plausibility check to differentiate between intended and unintended actions, providing a confidence score and safety decision-making to prevent misinterpretation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If AMRs operate with greater autonomy and closer human collaboration, then productivity and task capability are improved, but the risk of accidents and collisions increases

Engineering Contradiction:
Improvetask capabilityVSAvoidaccident risk
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system performs preliminary classification of human actions into intended and unintended categories before the AMR executes responses. The unintended action detector analyzes human behavior patterns, environmental context, and AMR state information in advance to identify potentially dangerous actions, allowing the system to prepare appropriate safety responses before actual collisions or accidents can occur.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary classification layer between human action detection and AMR response execution. This intermediary system (comprising the unintended action detector and plausibility checker) acts as a mediator that filters and validates human actions before they trigger AMR responses, preventing misinterpretation of unintended actions as intentional commands.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If conventional detection systems classify all human actions as intended, then ease of operation is maintained, but safety risks increase due to misinterpretation of unintended actions

Engineering Contradiction:
Improvedetection simplicityVSAvoidsafety risk
Core Design Contradiction:
Ease of operationVSObject-affected harmful factors

Solution Approach 1:

The detection system is segmented into distinct functional modules: an unintended action detector that performs initial classification, a plausibility checker that validates intended actions, and a safe decision-maker that executes responses. This segmentation allows each module to specialize in specific aspects of action analysis, improving both safety and operational ease.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The plausibility checker serves as an intermediary validation layer that verifies actions classified as intended by the detector. It checks whether the detected action aligns with environmental context, AMR state, and expected human behavior patterns, providing an additional safety filter without significantly increasing operational complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If the system implements comprehensive action analysis with multiple checks, then safety is improved, but device complexity increases

Engineering Contradiction:
ImprovesafetyVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system dynamically adjusts its analysis depth and computational resources based on the detected action type and environmental context. For high-risk situations or ambiguous actions, the system activates more thorough analysis protocols, while for clear-intended actions in safe contexts, it uses streamlined validation, optimizing the balance between safety and complexity.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes key parameters such as detection sensitivity thresholds, analysis depth levels, and response timing based on environmental context and AMR state. This allows the system to maintain high safety standards while adapting computational complexity to match the actual risk level of each situation.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4201608A1Unintended human action detection in an autonomous mobile robot environment
Publication Date: 2023.06.28 INTEL CORP
  • EP4201608A1 patent drawingFigure 1
  • EP4201608A1 patent drawingFigure 2
  • EP4201608A1 patent drawingFigure 3

AI summary

A computing device, including: a memory configured to store computer-readable instructions; and unintended human motion detection processing circuitry configured to execute the computer-readable instructions to cause the computing device to: interpret a human action; receive autonomous mobile robot (AMR) sensor data from an AMR sensor; and detect whether the human action is intended or unintended, wherein the detection is based on a predicted human action, a current human emotional or physical state, the interpreted human action, and the AMR sensor data.