Robot Manipulator Collision Detection Using Neural Network Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting collisions in robot manipulators, such as using torque sensors, are not economically viable for multi-axial robots and can increase system load due to the need for complex dynamics and kinematics calculations.
Innovation Solution
A system utilizing an artificial neural network that learns collision detection through data preprocessing, including cycle normalization, and uses joint actuators, encoders, and collision measurers like pressure sensors to detect collisions in real-time without the need for multiple torque sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If torque sensors are attached to each joint for collision detection, then collision detection sensitivity is improved, but system cost and complexity increase
Solution Approach 1:
The patent extracts the collision detection function from dedicated torque sensors and implements it using the existing motor current detection capability already present in the robot system. By utilizing the current measurement function that already exists in the motor controller, the system achieves collision detection without adding separate torque sensor hardware to each joint.
Solution Approach 2:
The motor current measurement system is made multi-functional by using it for both motor control and collision detection purposes. The same current sensors and processing circuits that serve the motor drive function are also utilized for detecting collision conditions, eliminating the need for separate dedicated detection hardware.
2Measurement precision
If dynamics and kinematics calculations are used for collision detection, then detection accuracy is improved, but system load increases
Solution Approach 1:
The patent extracts the essential collision detection capability from complex dynamics calculations and implements it through simpler motor current analysis. By focusing on the direct relationship between motor current and external forces, the system achieves accurate collision detection without requiring full dynamics and kinematics computations.
Solution Approach 2:
Instead of performing complete dynamics and kinematics calculations, the system applies partial action by using only the motor current measurement component that is most directly related to collision detection. This partial approach provides sufficient detection accuracy while significantly reducing computational load.
Data Source
AI summary
The present invention relates to a system for detecting the collision of a robot manipulator using an artificial neural network. The system may comprise: joint driving units provided in a plurality of joints of the robot manipulator to drive the plurality of joints, respectively; encoder units provided on sides of the joint driving units to measure the angles of the plurality of joints; and a neural network calculation unit for training the neural network with a large amount of data and inferring, via a preprocessing calculation, to detect that the plurality of joints collide with the outside.