Task Plan Generation Using Numerical Parameters and Search Trees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating task plans for autonomous systems are limited to operations with predefined symbols and cannot include operations with newly derived or arbitrary numerical values, leading to an inability to plan operations such as moving to an arbitrary location like a ⅔ point between two vertices.
Innovation Solution
The method involves constructing a causal operation network and a search tree based on a knowledge base that includes operation knowledge with specifications for using variables with arbitrary numerical values as parameters, allowing for the generation of task plans that include operations with these values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If symbols corresponding to all possible locations are individually defined and stored in the knowledge base in advance, then the task plan can include operations to arbitrary locations, but the number of symbols increases explosively making it unrealistic
Solution Approach 1:
The patent changes the parameter representation from discrete symbols to continuous numerical values. Instead of storing symbols for every possible location (e.g., '0_0', '10_0', '0_10'), the system uses numerical coordinates (e.g., 0, 10, 20) that can represent any position in the space. This allows arbitrary locations to be planned without exponentially increasing the symbol count, as numerical parameters can take any value within a range rather than requiring pre-defined discrete symbols.
2Productivity
If only predefined symbols are used in the knowledge base, then the task plan generation is computationally feasible, but operations with newly derived or arbitrary numerical values cannot be included
Solution Approach 1:
The system transitions from using fixed symbolic parameters to variable numerical parameters with discretization rules. The knowledge base stores operations with parameters that have expression types (integer, float, double) and discretization rules (minimum value, maximum value, value interval). This allows the system to generate task plans with arbitrary numerical values while maintaining computational feasibility through structured parameter management and discretization.
Solution Approach 2:
The patent introduces dynamic parameter generation where numerical values are derived during task plan execution rather than being fixed in advance. The system can calculate new numerical values on-the-fly (e.g., computing a location at 2/3 between two vertices) and use these dynamically generated values in operations, making the system adaptable to arbitrary locations without pre-defining all possible positions.
3Device complexity
If symbols are not treated as numbers and arithmetic operations are not performed during task plan generation, then the symbolic logical operations remain simple, but operations with values newly derived from individual unit information cannot be planned
Solution Approach 1:
The patent introduces dynamic arithmetic operations that are performed during task plan generation based on the current task requirements. Instead of statically defining all possible locations, the system can perform calculations (e.g., computing intermediate positions, determining distances, calculating trajectories) using the numerical parameters of objects and locations. This dynamic computation capability is triggered only when needed, maintaining simplicity for basic symbolic operations while enabling arithmetic when required.
Data Source
AI summary
The present invention relates to a method of generating a task plan, which includes constructing a causal operation network based on a knowledge base, constructing a search tree based on the knowledge base and the causal operation network, and generating a task plan based on the search tree.