Behavior Control System Using Stochastic Trajectory Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in maintaining continuity of agent behavior when switching between subtasks, leading to temporary operation stops, especially when executing a series of subtasks in a continuous manner.
Innovation Solution
A behavior control system that generates state variable trajectories using a stochastic transition model with a Dirichlet distribution to determine transition modes based on subtask execution probabilities, allowing the agent to selectively execute subtasks while ensuring continuity by optimizing joint probability trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a state variable trajectory is generated using a stochastic transition model with Dirichlet distribution, then the continuity of behavior is improved, but the arithmetic processing load increases
Solution Approach 1:
The patent segments the complex stochastic transition model into multiple subtask-specific models. Each subtask has its own transition model that can be independently evaluated, allowing the system to process only relevant portions of the state space at each decision point rather than evaluating all possible transitions simultaneously.
Solution Approach 2:
The patent pre-calculates and stores transition probabilities and Dirichlet distribution parameters for each subtask before execution. This preliminary preparation allows the runtime system to quickly retrieve and apply pre-computed values rather than performing complex arithmetic operations during real-time decision making.
2Productivity
If multiple subtasks are executed in a continuous manner, then the productivity is improved, but the difficulty of maintaining behavioral continuity increases
Solution Approach 1:
The patent implements a feedback mechanism where the current state variable value is continuously monitored and fed back into the stochastic transition model. This feedback loop allows the system to adjust transition probabilities based on the current state, ensuring that behavioral continuity is maintained even as the agent transitions between multiple subtasks at high speed.
Solution Approach 2:
The patent makes the transition mode dynamic by using the Dirichlet distribution to adaptively adjust transition probabilities based on the current state variable trajectory. This dynamic adjustment allows the system to maintain behavioral continuity while executing multiple subtasks continuously, as the transition behavior automatically adapts to the current execution context.
Data Source
AI summary
A system capable of causing an agent to continuously execute a plurality of different subtasks while securing the continuity of behavior of the agent is provided. A plurality of state variable trajectories representing the time series of a state variable of an object are generated according to a stochastic transition model in which the state variable of the object is represented as a random variable. The stochastic transition model is defined so that the transition mode of the state variable is determined according to an execution probability of each subtask in which a probability distribution is represented by a Dirichlet distribution. An operation of the agent is controlled so that the state of the object transits according to one state variable trajectory (desired state variable trajectory) maximizing or optimizing the joint probability of a whole of the stochastic transition model among the plurality of state variable trajectories.