Robot Controller Adaptation via Force Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic systems face challenges in efficiently adapting their trajectories in dynamic environments and interacting with humans, particularly in refining learned plans to accommodate unexpected changes through intuitive human input, as they struggle to accurately interpret noisy haptic data and adapt accordingly.
Innovation Solution
A robot device controller utilizing a statistical model, specifically a hidden semi-Markov model, to learn and adapt nominal task plans, combined with Bayesian optimization for online adaptation based on sensed interaction forces, allowing for efficient and intuitive trajectory adjustments through physical human-robot interaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the robot uses learning-from-demonstration to learn nominal task plans, then the robot can perform robotic tasks, but the learned plan needs refinement to adapt to unexpected environment changes and additional requirements
Solution Approach 1:
The system implements feedback by comparing expected forces (from nominal trajectory) with measured forces (from haptic sensors) and using this discrepancy to trigger adaptive refinement of the statistical model, enabling continuous improvement of trajectory planning based on actual execution experiences
Solution Approach 2:
The system adapts to environment changes by modifying parameters of the statistical model (which represents the nominal task plan) based on force discrepancies, allowing the robot to refine its understanding of task requirements without rewriting the entire control system architecture
2Ease of operation
If the robot physically interacts with human operator to show desired changes, then the robot can adapt trajectory intuitively, but the haptic data is noisy and requires accurate interpretation
Solution Approach 1:
The system uses the statistical model as an intermediary that translates noisy haptic measurements into meaningful trajectory refinements. The model acts as a buffer between the raw haptic data and the control actions, filtering noise while preserving intentional signals from the human operator
Solution Approach 2:
The system continuously compares expected forces from the nominal trajectory with actual measured forces, using this feedback loop to identify and respond to human-guided modifications while filtering out noise through the statistical model's learned patterns
3Speed
If the robot adapts trajectory quickly in response to haptic data, then the robot can respond to dynamic scenarios, but the adaptation must be accurate to fulfill additional requirements
Solution Approach 1:
The system performs preliminary adaptation by continuously monitoring force discrepancies and pre-adjusting the statistical model parameters before executing modified trajectories, allowing the robot to prepare for and smoothly implement trajectory changes without abrupt transitions
Solution Approach 2:
The system uses real-time feedback from force sensors to continuously refine the statistical model during task execution, enabling both rapid response to haptic inputs and accurate trajectory following through iterative comparison of expected versus measured forces
Data Source
Figure 1
Figure 2A~2B
Figure 3A~3B
AI summary
According to various embodiments, a robot device controller is described comprising a memory configured to store a statistical model trained to implement a behaviour of the robot device, one or more processors configured to determine a nominal trajectory represented by the statistical model, determine an expected force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory, determine a measured force experienced by the robot device when the robot device is controlled to move in accordance with the nominal trajectory and adapt the statistical model based on a reduction of the difference between the measured force and the expected force.