Autonomous Machine Pose Control Using Unified Quaternion Transforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for transforming coordinate frames in autonomous machines, such as robotics and autonomous vehicles, are complex and prone to errors due to the use of transformation matrices and dual quaternions, which can lead to bugs and inefficiencies in pose transformations.
Innovation Solution
A computing system that simplifies pose transformation by unifying pose and transformation using quaternions, allowing for easier mathematical addition of translations and rotations, thereby simplifying the transformation process and reducing errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If transformation matrix is used for coordinate frame transformation, then pose transformation can be achieved, but the transformation process becomes complex and prone to bugs
Solution Approach 1:
The patent changes the mathematical representation parameters from transformation matrices to quaternions. Quaternions use four parameters (w, x, y, z) instead of sixteen matrix elements, reducing complexity while maintaining transformation accuracy. The quaternion representation q = w + xi + yj + zk provides a compact way to encode both rotation and translation operations.
Solution Approach 2:
The patent extracts the essential transformation information (position and orientation) from complex transformation matrices and represents it using only the necessary quaternion parameters. This extraction eliminates redundant information and simplifies the transformation process while preserving the core functional capabilities.
2Reliability
If dual quaternions are used for pose transformation, then transformation can be achieved, but computational efficiency decreases and errors increase
Solution Approach 1:
The patent extracts only the essential quaternion components needed for pose transformation, discarding the redundant dual quaternion structure. By using simple quaternions instead of dual quaternions, the system maintains transformation accuracy while significantly improving computational efficiency and reducing error propagation.
Solution Approach 2:
The patent changes the parameter representation from dual quaternions (eight parameters) to simple quaternions (four parameters). This parameter reduction halves the computational burden while maintaining the ability to represent both rotation and translation operations effectively.
3Ease of operation
If algebraic method is used for inter-robot transformations, then transformation can be computed, but the method remains complex and transformations lack fixed sequence
Solution Approach 1:
The patent creates a universal quaternion-based transformation framework that can handle all coordinate frame transformations (inter-robot, intra-robot, sensor-to-world) in a unified manner. This universal approach eliminates the need for different algebraic methods for different transformation scenarios and establishes a fixed, consistent transformation sequence.
Data Source
AI summary
The present application discloses an autonomous machine and methods of controlling the autonomous machine. The present application also relates to a computing system for controlling movement of the autonomous machine. The autonomous machine comprises a movable component, a representation unit, a transformation unit communicatively coupled to the representation unit and a driving mechanism coupled to the movable component and the transformation unit for moving the movable component. In particular, poses of every part of the autonomous machine may be transformed back and forth by simply mathematical addition and subtraction.