Generative AI Warehouse Picking with Closed-Loop Robot Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing order-picking systems, particularly hybrid order-picking systems (HOPS), are brittle and require extensive training to handle complex scenarios, limiting their effectiveness and efficiency in warehouse operations.

Innovation Solution

A robot system utilizing a perceiver transformer and goal generator to autonomously navigate and pick objects in a warehouse, employing a perceiver transformer to generate latent vectors for end effector positioning and a goal generator to determine poses, combined with a planner for collision avoidance, enabling real-time closed-loop operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If hybrid order-picking systems (HOPS) are used to increase throughput, then productivity improves, but system reliability deteriorates due to brittleness

Engineering Contradiction:
ImprovethroughputVSAvoidsystem brittleness
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies parameter changes by transitioning from traditional rigid control parameters to generative AI-based probabilistic models. The system uses transformer architectures that generate actions based on learned distributions rather than fixed rules, allowing continuous adaptation to environmental variations while maintaining high throughput capabilities.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system implements dynamics by enabling real-time adaptation through continuous learning from sensor feedback. The generative AI models dynamically adjust their behavior based on changing warehouse conditions, object locations, and environmental factors, transforming the brittle static system into a flexible adaptive one that maintains productivity.

Inventive Principle:
Principle #15Dynamics

2Reliability

If extensive training is provided to handle complex scenarios, then system reliability improves, but device complexity increases

Engineering Contradiction:
Improvehandling complex scenariosVSAvoidtraining requirements
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system applies self-service through autonomous continuous learning capabilities. The generative AI models automatically improve their performance by processing new data and feedback from the environment without requiring manual retraining or intervention. This enables the system to handle complex scenarios through self-improved adaptability rather than extensive pre-programming.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent implements feedback mechanisms where the system continuously monitors its performance and uses this information to refine its generative models. The feedback loop allows the system to learn from successes and failures in real-time, improving reliability for complex scenarios through iterative optimization rather than static training.

Inventive Principle:
Principle #23Feedback

3Productivity

If autonomous systems are deployed to reduce cost per pick, then productivity improves, but ease of operation deteriorates due to training requirements

Engineering Contradiction:
Improvecost per pickVSAvoidtraining requirements
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system achieves self-service operation through generative AI that autonomously handles picking tasks without human intervention. The models self-correct errors and adapt to new situations independently, eliminating the need for extensive human training while maintaining operational flexibility and reducing cost per pick through efficient autonomous execution.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4578606A1Technologies for generative ai for warehouse picking
Publication Date: 2025.07.02 INTEL CORP
  • EP4578606A1 patent drawingFigure 1
  • EP4578606A1 patent drawingFigure 2
  • EP4578606A1 patent drawingFigure 3

AI summary

Technologies for generative AI for warehouse picking are disclosed. In an illustrative embodiment, a robot can partially or fully autonomously perform object picking in a warehouse. A description of an object to be picked can be sent to a compute device controlling the robot. The compute device can direct the robot to move to where the object is located. The robot can then take a picture that can be analyzed. A text description as well as the image is encoded and provided to a transformer. The transformer generates an output vector indicating, e.g., where an end effector should grab hold of an object. A goal generator can then determine a pose for the end effector based on the output latent vector. A planner can move the end effector to the determined pose. A new picture can be taken, and the cycle can repeat until the item is picked.