Robot Base Position Planning Using Height Maps and ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robotic systems face inefficiencies in determining collision-free trajectories for their base position when interacting with physical environments, leading to slow operation and hesitation, as they rely on iterative guess-and-check methods that are computationally expensive and time-consuming.
Innovation Solution
A pre-trained machine learning model, such as an artificial neural network, is used to predict candidate positions for the robotic device's base based on a height map of the environment, allowing the manipulator to reach interaction points without collisions, significantly reducing the time required to find a suitable position.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If iterative guess-and-check methods are used to determine collision-free trajectories, then the robot can ensure safety by validating each candidate position, but the operation speed decreases and the robot appears hesitant
Solution Approach 1:
The system performs preliminary action by using a pre-trained machine learning model to predict candidate base positions that are likely to be collision-free before the robot actually moves. This allows the robot to have pre-computed trajectory options ready, reducing the need for slow iterative validation during real-time operation while maintaining safety through subsequent verification of the predicted positions.
2Measurement precision
If conventional iterative methods are used to find candidate base positions, then the robot can determine valid positions, but the time required to perform tasks increases
Solution Approach 1:
The system replaces the mechanical iterative guess-and-check method with a machine learning-based predictive system. The pre-trained model processes the height map and interaction point to directly generate candidate base positions, substituting the slow computational iteration with a faster learning-based inference process that maintains accuracy in position determination.
3Productivity
If the robot uses pre-trained machine learning models to predict candidate positions, then the operational efficiency increases, but the system complexity increases due to model training and integration
Solution Approach 1:
The system addresses complexity by performing the computationally intensive model training in advance, before deployment. The pre-trained model is stored and reused during operation, converting the complexity from an ongoing operational burden to a one-time setup cost. This allows the robot to benefit from fast inference during actual tasks without bearing the continuous computational load of model training.
Data Source
AI summary
A method includes receiving sensor data representative of surfaces in a physical environment containing an interaction point for a robotic device and determining, based on the sensor data, a height map of the surfaces in the physical environment. The method also includes determining, by inputting the height map and the interaction point into a pre-trained model, one or more candidate positions for a base of the robotic device to allow a manipulator of the robotic device to reach the interaction point. The method additionally includes determining a collision-free trajectory to be followed by the manipulator of the robotic device to reach the interaction point when the base of the robotic device is positioned at a selected candidate position of the one or more candidate positions and, based on determining the collision-free trajectory, causing the base of the robotic device to move to the selected candidate position within the physical environment.