Robot Arm Operation Learning for Faster, Low-Shake Spot Welding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging for human operators to set optimal trajectories, speeds, and accelerations for robot arms in spot welding robots to achieve short cycle times while minimizing shaking and ensuring accurate welding, as manually determining the shortest path and adjusting for variations in arm movement is difficult.
Innovation Solution
A machine learning device that observes arm shaking and trajectory length, calculates cycle time, and learns an operation program using reinforcement learning to optimize robot movements, incorporating neural networks and camera data to adjust speed and acceleration for improved efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a teacher manually sets teaching points and trajectories for the robot arm to pass through multiple welding points, then the robot can perform spot welding operations, but it is difficult to achieve the shortest trajectory and optimal cycle time due to the complexity of manual settings
Solution Approach 1:
The robot system performs self-learning through reinforcement learning, where the robot autonomously optimizes its own operation program by trial and error. The learning unit enables the robot to automatically find optimal trajectories and operation sequences without human intervention, allowing the system to serve itself in optimizing its performance.
Solution Approach 2:
The patent replaces manual teaching methods with an automated machine learning system. Instead of a teacher manually setting teaching points and trajectories, the system uses reinforcement learning algorithms to automatically optimize the robot's operation program, substituting mechanical manual adjustment with computational optimization.
2Productivity
If the robot arm moves faster to reduce cycle time, then productivity improves, but shaking of the robot arm increases which reduces welding accuracy
Solution Approach 1:
The operation program is optimized dynamically to balance speed and accuracy. The reinforcement learning process learns optimal speed profiles and acceleration patterns that minimize cycle time while keeping shaking within acceptable limits for accurate welding, allowing dynamic adjustment of movement parameters.
Solution Approach 2:
The system optimizes multiple parameters simultaneously including trajectory points, speeds, and accelerations. By changing these parameters through reinforcement learning, the system finds the optimal combination that achieves short cycle times while maintaining welding accuracy through controlled shaking levels.
3Manufacturing precision
If the robot arm moves slower to reduce shaking and improve welding accuracy, then manufacturing precision improves, but cycle time increases reducing productivity
Solution Approach 1:
The system uses dynamic movement control with optimized acceleration and deceleration patterns. Rather than uniformly slow movement, the robot performs rapid movements when accuracy is not critical and slows down only when necessary for precise welding operations, achieving both speed and accuracy through dynamic parameter adjustment.
4Ease of operation
If multiple teaching points are set between welding points to optimize trajectory, then the robot can achieve better path control, but the complexity of manual teaching settings increases significantly
Solution Approach 1:
The robot autonomously determines optimal teaching points and trajectories through self-learning. Instead of requiring a teacher to manually set multiple teaching points, the reinforcement learning system automatically identifies optimal path points and generates the teaching program, eliminating the complexity of manual trajectory planning.
Solution Approach 2:
The system uses simulation environments to copy and practice welding operations before executing them on the actual robot. This allows the robot to learn optimal trajectories and teaching points in a virtual environment, then transfer this knowledge to the physical system, avoiding complex manual teaching procedures.
Data Source
AI summary
A machine learning device, which learns an operation program of a robot, includes a state observation unit which observes as a state variable at least one of a shaking of an arm of the robot and a length of an operation trajectory of the arm of the robot; a determination data obtaining unit which obtains as determination data a cycle time in which the robot performs processing; and a learning unit which learns the operation program of the robot based on an output of the state observation unit and an output of the determination data obtaining unit.