Autonomous Connector Mating Using Vision-Guided Maneuver Generation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current high-speed docking operations, such as air-to-air refueling and spacecraft docking, heavily rely on human judgment and are complicated by factors like relative motion, poor visibility, and inaccurate depth perception, leading to potential damage during connector mating.

Innovation Solution

A system utilizing a trained autonomous agent, such as a neural network, to generate maneuvering recommendations for connector mating based on image and position data, improving reliability and standardization by processing 2D image data to determine 3D coordinates and using reinforcement learning to optimize maneuvers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If human operators guide complex high-speed docking operations, then the operations can be performed with judgment and adaptability, but the operations are slow and prone to error due to poor visibility and inaccurate depth perception

Engineering Contradiction:
Improvedocking operation reliabilityVSAvoiddocking operation speed
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent replaces human operators with an autonomous system that uses computer vision and reinforcement learning to guide docking operations. The trained agent processes sensor data and generates control commands automatically, eliminating the need for human judgment while maintaining high reliability through learned policies from simulated training.

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

Solution Approach 2:

The patent introduces a trained autonomous agent as an intermediary between sensor data and control commands. This agent, trained through reinforcement learning in simulation, acts as a mediator that translates raw sensor inputs into reliable docking maneuvers, improving both reliability and speed by removing human reaction time limitations.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If complex stereoscopic vision systems are used to aid human operators, then depth perception is improved, but the system complexity and cost increase

Engineering Contradiction:
Improvedepth perception accuracyVSAvoidvision system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex stereoscopic vision systems with a simplified monocular camera system. The trained autonomous agent, through reinforcement learning, learns to infer depth and spatial relationships from 2D image data, eliminating the need for expensive and complex stereoscopic hardware while achieving comparable or superior measurement precision.

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

Solution Approach 2:

The patent changes the approach from hardware-based depth perception (stereoscopic cameras) to software-based depth inference (reinforcement learning agent processing 2D images). This parameter change from optical to computational methods reduces device complexity while maintaining or improving depth perception accuracy through learned features.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If traditional computer vision techniques are used for connector mating, then the system can process visual data, but the maneuver execution rate is limited and reliability is insufficient

Engineering Contradiction:
Improvemaneuver execution rateVSAvoidconnector mating reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary training of the autonomous agent in simulated environments before real-world deployment. Through extensive reinforcement learning training in simulation, the agent learns optimal docking policies that can be rapidly executed in real operations, achieving both high maneuver execution rates and reliability by pre-learning from synthetic data.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements a feedback loop where the autonomous agent receives continuous sensor data during docking operations and adjusts its maneuvers in real-time. The reinforcement learning framework provides feedback through reward signals during training, enabling the agent to learn corrective actions that improve both execution rate and reliability during actual connector mating operations.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4053020B1Autonomous maneuver generation to mate connectors
Publication Date: 2024.10.16 THE BOEING CO
  • EP4053020B1 patent drawingFigure 1
  • EP4053020B1 patent drawingFigure 2~3
  • EP4053020B1 patent drawingFigure 4A~4D

AI summary

A method includes providing an image to a feature extraction model to generate feature data. The image depicts a portion of a first device and a portion of a second device. The feature data includes coordinates representing key points of each of the first and second devices depicted in the image. The method also includes obtaining position data indicating a position in 3D space of a connector of the first device. The method further includes providing the feature data and the position data to a trained autonomous agent to generate a proposed maneuver to mate the connector with a connector of the second device.