Robotic Object Placement Using Vision-Guided Motion Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Robots struggle to adapt to dynamic or varying environments when placing secured objects in specific locations without explicit user programming, often failing if the target location changes or if the environment is unfamiliar.

Innovation Solution

A machine learning model, such as a deep neural network, processes images of the secured object and its environment to predict the likelihood of successful placement, adjusting end effector motions iteratively to reach the target location and release the object when criteria are met, using training examples from real and simulated data to enhance adaptability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a robot is programmed to place objects in a preprogrammed fixed location, then the placement operation is simple and reliable, but the robot fails when the bin is moved to a different location or in a new environment

Engineering Contradiction:
Improveplacement success rateVSAvoidenvironment adaptability
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system transitions from static preprogrammed locations to dynamic adaptive location determination. The machine learning model continuously learns from environmental feedback and adjusts placement locations dynamically, allowing the robot to adapt to moved bins and new environments while maintaining high placement success rates

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system implements feedback loops where the robot observes the results of placement attempts and uses this information to improve future placements. The machine learning model is trained on collected data from real and simulated environments, creating a feedback mechanism that enhances adaptability without requiring explicit reprogramming

Inventive Principle:
Principle #23Feedback

2Reliability

If a robot uses explicit user programming for placement locations, then the placement operation is precise and reliable, but the robot cannot adapt to new environments without reprogramming

Engineering Contradiction:
Improveplacement precisionVSAvoidprogramming complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The robot performs self-learning and self-programming through the machine learning model. Instead of requiring explicit user programming for each new environment, the system automatically adapts by learning from observed data, reducing programming complexity while maintaining placement precision through automated environmental understanding

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system changes the approach from fixed programmed parameters to learned parameters that adapt to environmental conditions. The machine learning model adjusts placement parameters dynamically based on learned patterns from training data, maintaining precision while eliminating the need for complex reprogramming

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If a robot uses traditional programming for dynamic environments, then the robot can handle varying conditions, but the system becomes complex and requires frequent reprogramming

Engineering Contradiction:
Improveenvironment flexibilityVSAvoidcontrol system complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical control systems with machine learning-based intelligent control. Instead of complex if-then programming rules, the system uses neural networks to automatically determine placement strategies, reducing control system complexity while enhancing environmental flexibility through learned behavioral patterns

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

Data Source

PatentEP3867020B1Machine learning methods and apparatus for automated robotic placement of secured object in appropriate location
Publication Date: 2025.08.06 GDM HOLDING LLC
  • EP3867020B1 patent drawingFigure 1
  • EP3867020B1 patent drawingFigure 2
  • EP3867020B1 patent drawingFigure 3

AI summary

Training and/or use of a machine learning model for placement of an object secured by an end effector. A machine learning model can be used to process: (1) a current image, captured by a robot vision component, that captures an end effector securing an object; (2) a candidate end effector action that defines a candidate motion of the end effector; and (3) a target placement input that indicates a target placement location for the object. Based on the processing, a prediction can be generated that indicates likelihood of successful placement of the object in the target placement location with application of the motion defined by the candidate end effector action. At many iterations, the candidate end effector action with the highest probability is selected and corresponding control commands provided. When release criteria/criterion are satisfied, control commands can be provided to cause the end effector to release the object.