Robot Pose Calibration Using Camera Trajectory Curve Fitting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing robot positioning methods based on labels are prone to significant errors due to changes in robot setup during daily use, leading to inaccuracies in pose information determination.

Innovation Solution

A method and apparatus that determine pose information by fitting a target curve to the motion trajectory of an image module relative to a target label, calculating coordinate offsets, and determining relative distances to accurately assess the robot's pose, even with offsets between the image module and the robot.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If robot positioning is based on label acquisition using robot camera, then positioning can be achieved, but positioning accuracy deteriorates when offset occurs between camera and robot body

Engineering Contradiction:
Improvepositioning accuracyVSAvoidpositioning reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary action by acquiring the motion trajectory of the image module before actual positioning operations. By pre-capturing the trajectory data when the robot moves to a known position, the system establishes a reference curve that accounts for any installation offsets. This preliminary measurement enables subsequent positioning operations to compensate for camera-robot body misalignment, thereby maintaining high positioning accuracy and reliability.

Inventive Principle:
Principle #10Preliminary action

2Ease of operation

If camera is installed on robot body, then robot can acquire label position, but installation offsets cause errors in pose information determination

Engineering Contradiction:
Improveoperation capabilityVSAvoidpose information accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The system implements feedback by using the acquired motion trajectory to generate a reference curve, then comparing actual camera positions against this curve. The trajectory data feedback allows the system to calculate accurate pose information by referencing the pre-established motion pattern, compensating for any installation offsets between the camera and robot body, thus maintaining measurement precision while preserving operational ease.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If robot moves to known position for positioning, then pose information can be determined, but installation offsets cause large errors

Engineering Contradiction:
Improvepose determination accuracyVSAvoidpositioning system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system introduces an intermediary element - the motion trajectory curve - that mediates between the robot's physical movement and the camera's position measurements. By using this trajectory curve as an intermediate reference, the system can determine accurate pose information without directly relying on precise camera-robot body alignment, thereby improving measurement precision while avoiding the complexity of complex calibration systems.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS11644302B2Method and apparatus for determining pose information of a robot, device and medium
Publication Date: 2023.05.09 KEENON ROBOTICS CO LTD
  • US11644302B2 patent drawing
  • US11644302B2 patent drawing
  • US11644302B2 patent drawing

AI summary

Provided are a method and apparatus for determining pose information of a robot, a device and a medium. The method includes: fitting a target curve according to a motion trajectory of an image module relative to a target label when the robot rotates; determining a coordinate offset of the image module relative to a body of the robot according to the target curve, and determining a relative distance between the image module and the target label according to the image module; and determining the pose information of the robot according to the coordinate offset and the relative distance.