Virtual Character Action Decisions Using Fused Status and Environment Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining virtual character actions in complex virtual environments consume significant resources and labor due to the large scale and complexity of the environments, leading to high computational complexity and twitching issues.
Innovation Solution
An action decision-making method that fuses character status information and multimodal environmental perception information using different coding schemes to generate a fusion feature, enabling more accurate and efficient decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image recognition is performed on raw pixel images to determine virtual environment information, then comprehensive environmental perception is achieved, but computational resources and time consumption increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-processing environmental information into structured data formats before decision-making occurs. The game environment is pre-analyzed to extract key elements (obstacles, targets, pathways) and represent them in optimized data structures, so that during runtime, the neural network receives pre-processed information rather than raw pixel data, significantly reducing real-time computational burden while maintaining perception accuracy
Solution Approach 2:
The patent extracts essential environmental features from complex game scenes by identifying and isolating key elements such as obstacles, targets, and navigable pathways. This extraction process removes unnecessary visual noise and focuses computational resources on critical decision-making information, reducing the data volume fed to the neural network while preserving actionable environmental intelligence
2Reliability
If comprehensive environmental information is captured and processed, then action decision accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments environmental information into distinct categorical components (obstacles, targets, pathways, hazards) that can be processed independently by specialized neural network modules. Each segment is encoded separately using modality-specific representations, allowing parallel processing and reducing overall computational complexity while maintaining comprehensive environmental understanding for accurate decision-making
Solution Approach 2:
The patent transforms environmental data from raw visual representations into optimized parameterized formats suitable for neural network input. By converting game state information into structured numerical representations with appropriate scaling and normalization, the system reduces computational complexity of processing while preserving all necessary information for accurate action decision-making
3Measurement precision
If feature coding is performed on character status and environmental perception information, then decision-making accuracy improves, but processing time increases
Solution Approach 1:
The patent merges character status information and environmental perception information into a unified feature representation that is processed simultaneously by the neural network. By combining these data streams into a single integrated input tensor with appropriate channel organization, the system eliminates sequential processing overhead and enables parallel computation of both character state and environmental context, improving processing speed while maintaining high decision-making accuracy
Data Source
AI summary
An action decision-making method for a virtual character is performed by a computer device. The method includes: obtaining character status information of a first virtual character and environmental perception information of the first virtual character with respect to a virtual environment in which the first virtual character is currently located; fusing the character status information and the environmental perception information, to obtain a fusion feature; determining a character action for the first virtual character by applying the fusion feature to an action decision-making model; and controlling the first virtual character to execute the character action.