Robot Task Programming via Language Model Subtask Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robotic systems require extensive manual reprogramming and expert knowledge to adapt to new tasks, making them inefficient in performing modified assembly processes.
Innovation Solution
A computer-implemented method that uses two machine learning models to generate program code for robots. The method receives user input specifying a task, processes it to generate subtasks, and then processes each subtask to produce executable code for controlling the robot.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional manual reprogramming methods are used, then the robot can be controlled to perform tasks with high precision, but the system requires extensive time and expert knowledge for adaptation to new tasks
Solution Approach 1:
The patent introduces an intermediary system consisting of a language model and code generator that translates natural language task descriptions into executable robot code. This intermediary layer eliminates the need for direct manual programming while maintaining task execution precision, as the generated code is then validated through simulation before deployment.
Solution Approach 2:
The system performs preliminary actions by generating and validating robot code through simulation environments before actual execution. This preliminary validation process ensures task precision is maintained while reducing reprogramming time, as errors are caught in simulation rather than during real task execution.
2Measurement precision
If conventional manual reprogramming methods are used, then the robot can be controlled to perform tasks with high precision, but the system requires considerable engineering expertise for task adaptation
Solution Approach 1:
The language model acts as an intermediary that translates natural language descriptions into precise robot control code, eliminating the need for users to have engineering expertise in robot kinematics or control programming languages while maintaining task execution precision through the code generation and simulation validation process.
Solution Approach 2:
The system enables self-service by allowing users to describe tasks in natural language without requiring expert knowledge of robot programming. The automated code generation and simulation validation processes handle the technical complexity, making task adaptation accessible to non-experts while maintaining high precision.
3Productivity
If automated code generation is used, then the robot can adapt to new tasks quickly without expert knowledge, but the system requires validation to ensure correct task execution
Solution Approach 1:
The system creates a virtual copy of the robot environment through simulation, where generated code is validated before actual execution. This copying approach enables rapid task adaptation while managing validation complexity by testing in a virtual environment that mirrors the physical system without the risks of direct real-world testing.
Solution Approach 2:
Validation through simulation is performed as a preliminary action before actual task execution. This approach enables quick task adaptation by catching errors in the virtual environment, while the structured simulation process manages complexity by providing a controlled testing framework that validates code before deployment.
Data Source
AI summary
One embodiment of a method for generating program code to control a robot includes receiving user input specifying a task to be performed by the robot, processing the user input via a first machine learning model to generate a plurality of subtasks for performing the task, and for each subtask included in the plurality of subtasks, processing the subtask via a second machine learning model to generate program code for controlling the robot to perform the subtask.