Robotic Item Handler Using 3D Point Cloud ML for Mode Selection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Manual loading and unloading of vehicles at material handling sites is costly and inefficient, necessitating an improved automated system for quickly and effectively handling bulk quantities of stacked or scattered items.

Innovation Solution

A robotic item handler equipped with multiple sensors and a machine learning model that determines the optimal operating mode, either grasping or sweeping, based on three-dimensional image data to efficiently load or unload items.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated loading/unloading systems are implemented, then productivity and labor cost efficiency are improved, but device complexity and initial investment cost increase

Engineering Contradiction:
Improveloading and unloading efficiencyVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The robotic item handler is divided into distinct functional modules: a robotic arm for grasping operations, a platform for sweeping operations, multiple sensor devices for perception, and a processing unit for decision-making. Each module operates semi-independently, allowing for easier maintenance and reduced overall system complexity while maintaining high productivity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system dynamically switches between different operating modes (grasping vs. sweeping) based on real-time analysis of three-dimensional image data. The processing unit evaluates the spatial arrangement of items and automatically selects the most efficient operating mode, enabling the system to adapt to varying loading conditions and maximize productivity.

Inventive Principle:
Principle #15Dynamics

2Adaptability or versatility

If multiple sensor devices and machine learning models are used, then decision accuracy and adaptability are improved, but device complexity and processing requirements increase

Engineering Contradiction:
Improveoperating mode selection capabilityVSAvoidcontrol system complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The control system replaces complex mechanical decision-making with a machine learning model that processes three-dimensional image data from multiple sensor devices. The model automatically determines the optimal operating mode based on learned patterns from training data, simplifying the control architecture while enhancing adaptability to different item configurations.

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

Solution Approach 2:

The machine learning model is trained offline using labeled training data, enabling the system to make autonomous decisions during operation without requiring real-time human intervention. The model self-adjusts by evaluating features such as item stacking patterns, spatial distribution, and container geometry to independently select the most appropriate operating mode.

Inventive Principle:
Principle #25Self-service

3Reliability

If machine learning models are trained with extensive data, then decision accuracy is improved, but processing time and computational resources increase

Engineering Contradiction:
Improvedecision classification accuracyVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The machine learning model is trained in advance using extensive labeled training data that includes various item configurations, stacking patterns, and container types. This preliminary training phase allows the model to learn optimal decision-making strategies offline, so that during actual operation, the model can quickly classify operating modes with high accuracy without requiring extensive real-time processing.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4706915A2Machine learning based decision making for robotic item handling
Publication Date: 2026.03.11 INTELLIGRATED HEADQUARTERS LLC
  • EP4706915A2 patent drawingFigure 1
  • EP4706915A2 patent drawingFigure 2
  • EP4706915A2 patent drawingFigure 3

AI summary

A method for controlling a robotic item handler is described. The method includes obtaining first point cloud data related to a first three-dimensional (3D) image and second point cloud data related to a second 3D image, captured by a first sensor device and a second sensor device respectively. Further, the method can include transforming the first point cloud data and the second point cloud data to combined point cloud data that is used as an input to a convolutional neural network, to construct a machine learning model. The machine learning model can output a decision classification indicative of a first probability associated with a first operating mode and a second probability associated with a second operating mode. Furthermore, the method can include operating the robotic item handler according to the first operating mode or the second operating mode based on a comparison of the first probability and the second probability.