3D Trajectory Post-Correction Using Curvature Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current direct teaching methods for industrial robots are prone to noise and inaccuracies due to user hand shaking and sensor/electronic circuit noise, leading to distorted trajectories and reduced accuracy in reproducing the taught motion.
Innovation Solution
A method for post-correction of 3D feature point-based direct teaching trajectories that extracts shape-based feature points using curvature and velocity, and improves the correction algorithm to classify segments as lines or curves, effectively removing noise and enhancing accuracy by using a torque sensor and preprocessing steps like the Douglas-Peucker algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If direct teaching is performed by manually moving the robot along the trajectory, then the robot can be taught the desired motion path, but the recorded trajectory contains noise and inaccuracies due to user hand shaking and sensor errors
Solution Approach 1:
The patent applies preliminary action by performing preprocessing steps (Douglas-Peucker algorithm for removing excessive measurement values, line-smoothing for high-frequency noise) and post-correction steps (feature point extraction, segment classification, trajectory reconstruction) before the robot executes the taught trajectory. This ensures that noise and inaccuracies are eliminated in advance, allowing the robot to reproduce the trajectory with high precision without requiring perfect manual teaching execution
2Measurement precision
If a line-smoothing algorithm is used to handle high-frequency noise, then the trajectory can be corrected to some extent, but low-frequency noise from user hand shaking and trajectory distortion at corners remain problematic
Solution Approach 1:
The patent applies segmentation by dividing the trajectory into multiple segments based on feature points (extracted using curvature and velocity criteria). Each segment is independently classified as either a line or curve type, and corrected separately using appropriate methods. This segmented approach allows the system to handle different types of trajectory portions differently, effectively removing both high-frequency noise and low-frequency distortion while maintaining corner accuracy and avoiding excessive complexity in a single monolithic algorithm
3Productivity
If the robot reproduces the recorded trajectory with high velocity, then productivity is improved, but high-frequency noise causes abnormal operations requiring excessive velocity
Solution Approach 1:
The patent applies preliminary action by performing comprehensive noise filtering and trajectory correction before the robot executes the motion. The Douglas-Peucker algorithm removes excessive measurement values, line-smoothing eliminates high-frequency noise, and post-correction reconstructs the trajectory using feature points. This preliminary processing ensures that the trajectory is clean and accurate, allowing the robot to reproduce it at high velocity without abnormal operations or excessive velocity requirements, thus improving both productivity and reliability
Data Source
AI summary
There is provided a method of post-correction of a 3D feature point-based direct teaching trajectory, which improves direct teaching performance by extracting shape-based feature points based on curvature and velocity and improving a direct teaching trajectory correction algorithm using the shape-based feature points. Particularly, there is provided a method of post-correction of a 3D feature point-based direct teaching trajectory, which makes it possible to extract and post-correct a 3D (i.e., spatial) trajectory, as well as a 2D (i.e., planar) trajectory, with higher accuracy.