Humanoid Robot Behavior Planning for Close Physical Interaction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robots lack the ability to effectively handle emergencies and emergencies in an open dynamic environment and autonomous development and adaptability in close interaction with humans, particularly in environments where they struggle to understand their own behaviors and predictably interact with humans, and their ability to perform physical tasks like hugging and dexterous grasping.
Innovation Solution
A method for planning and controlling a robot to understand its own behaviors and predict the consequences of physical interactions by using motion capture data, clustering analysis, and a hierarchical directed graph to plan and execute humanoid behaviors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If robots use pre-programming for automation, then automation capability is improved, but ability to handle emergencies and adapt to dynamic environments deteriorates
Solution Approach 1:
The patent implements a dynamic behavior planning system that transitions from static pre-programming to real-time adaptive decision-making. The robot uses a hierarchical directed graph structure with movement primitives that can be dynamically selected and combined based on current environmental conditions and perceived human intentions, enabling both automation and adaptability to emergencies.
Solution Approach 2:
The patent incorporates perception modules that continuously monitor environmental conditions and human behaviors, feeding this information back to the behavior planning system. This feedback loop enables the robot to adjust its automated actions in real-time based on dynamic environmental changes and emergent situations, resolving the contradiction between automation and emergency handling capability.
2Measurement precision
If robots rely on large amounts of data for learning, then behavior understanding accuracy is improved, but autonomous development capability and task universality deteriorate
Solution Approach 1:
The patent segments complex human-robot interaction behaviors into atomic movement primitives that can be independently learned and recombined. By dividing behavior into discrete, reusable units with specific functions and parameters, the system achieves high behavior understanding accuracy through focused learning while maintaining task universality through flexible combination of primitives.
Solution Approach 2:
The patent creates a universal behavior framework where movement primitives serve multiple functions across different tasks. Each primitive is designed with adjustable parameters that allow it to be applied in various contexts, enabling the robot to generalize from limited data while maintaining accurate behavior understanding across diverse interaction scenarios.
3Productivity
If robots perform close physical interactions, then human-robot collaboration quality is improved, but ability to understand own behaviors and predict perception consequences deteriorates
Solution Approach 1:
The patent implements self-perception feedback mechanisms where the robot monitors its own executed actions and compares them with expected outcomes. This internal feedback loop enables the robot to understand its own behaviors by tracking action execution and predicting perception consequences, maintaining both high collaboration quality and self-awareness during close physical interactions.
Solution Approach 2:
The patent introduces a behavior planning layer that acts as an intermediary between action execution and perception outcomes. This intermediate representation layer provides a structured model of intended actions and expected consequences, enabling the robot to predict perception outcomes before execution and understand its own behaviors during close physical interactions.
Data Source
AI summary
The present disclosure relates to a method for planning and controlling a humanoid behavior of a robot for close physical interactions, which includes: step S1, acquiring demonstration data of close interactions between the robot and human beings through motion capture; step S2, obtaining a plurality of human behavior pattern categories based on the demonstration data by using prior knowledge and clustering analysis; step S3, segmenting and calibrating the demonstration data based on the plurality of human behavior pattern categories to obtain a plurality of groups of movement primitive sequences including human behavior pattern labels, constructing a hierarchical directed graph, and obtaining a robot behavior planner through training; and step S4, constructing a dynamically consistent mapping model between a target trajectory and an action space, and realizing the planning and control of the humanoid behaviors of the robot based on the dynamically consistent mapping model and the robot behavior planner.