Robot Motion Trajectory Prediction Without RCS Modules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic motion simulation techniques, especially for robots without Robot Controller Software (RCS) modules, face challenges in accuracy and cost due to reliance on generic motion planners that may not support all robotic kinematics and result in high licensing costs and slow communication.
Innovation Solution
A method using machine learning to predict motion trajectories by processing data from various sources, including physical and simulated robot motions, to generate a motion prediction module that can simulate robots with complex kinematics without requiring RCS modules or external client-server communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If generic motion planners are used for robots without RCS modules, then licensing costs are reduced, but motion prediction accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by automatically generating motion training data through virtual simulations before the actual motion prediction is needed. This pre-computed training data enables the machine learning model to achieve high prediction accuracy without requiring complex manual configuration of motion parameters during runtime.
Solution Approach 2:
The system creates a virtual copy of the physical robot in a simulated environment to generate training data. This virtual robot copy allows the system to learn motion patterns from simulated operations without requiring access to the actual robot's proprietary RCS modules, thereby maintaining accuracy while avoiding licensing costs.
2Reliability
If proprietary RCS modules are used, then motion prediction accuracy is improved, but licensing costs increase
Solution Approach 1:
The system replaces expensive proprietary RCS modules with a cost-effective machine learning-based prediction system. The ML model, trained on virtual simulation data, provides accurate motion predictions without requiring ongoing licensing fees for proprietary robot controller software.
Solution Approach 2:
The system substitutes the mechanical/proprietary RCS module-based motion prediction approach with a software-based machine learning system. This substitution eliminates dependency on vendor-specific proprietary software while maintaining prediction accuracy through data-driven learning.
3Reliability
If proprietary RCS modules are used, then motion prediction accuracy is improved, but communication speed deteriorates
Solution Approach 1:
The system extracts the essential motion prediction functionality from the proprietary RCS modules and implements it as a standalone machine learning system. This extraction eliminates the need for slow external client-server communication with proprietary modules, enabling fast local prediction while retaining accuracy.
4Adaptability or versatility
If manual configuration of motion parameters is required, then adaptability to specific robots is improved, but ease of operation deteriorates
Solution Approach 1:
The system performs self-service by automatically identifying robot characteristics and adapting to different robot types without requiring manual configuration. The machine learning model learns robot-specific motion patterns directly from virtual simulation data, eliminating the need for users to manually input motion parameters or configure kinematics settings.
Solution Approach 2:
The system achieves universality by designing a unified machine learning framework that can handle multiple robot types and kinematics configurations through a single automated training process. This universal approach supports various robot platforms without requiring separate manual configuration procedures for each robot type.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and a method for predicting a motion trajectory of a robot moving between a given pair of robotic locations. Training data of motion trajectories of the robot are received for a plurality of robotic location pairs. The training data are processed so as to obtain x tuples and y tuples for machine learning purposes; wherein the x tuples describe the robotic location pair and the y tuples describe one or more intermediate robotic locations at specific time stamps during the motion of the robot between the locations of the location pair. From the processed data, a function is learned for mapping the x tuples into the y tuples so as to generate a motion prediction module for the robot. For a given robotic location pair, the robotic motion between the given pair is predicted by obtaining the corresponding intermediate locations resulting from the motion prediction module.