Multi-Agent Path Planning With Global Time-Aware Route Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing path planning methods, such as D* and DWA, are inadequate in handling multi-body dynamic environments due to the lack of consideration for the influence of other autonomous mobile agents and limited time windows, making them impractical for real-world scenarios.
Innovation Solution
A path planning method that integrates static and dynamic road network information to model the environment, including motion states of multiple agents, using a feature extraction network and a value estimation network to determine optimal paths, incorporating Graph-Bert and LSTM for environmental feature extraction and prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If single-body dynamic planning methods (e.g., D*) are used, then the path planning can be implemented, but the performance is poor in multi-body dynamic environments due to lack of consideration for influences of other autonomous mobile agents
Solution Approach 1:
The patent segments the complex multi-body dynamic environment into individual agent models, each with its own motion characteristics and constraints. By modeling each agent separately and considering their interactions, the system can handle multi-body dynamics effectively while maintaining reliable path planning performance.
Solution Approach 2:
The patent implements feedback mechanisms where the path planning system continuously monitors the states and movements of multiple agents, and adjusts the planned paths in real-time based on the actual environmental feedback. This allows the system to adapt to dynamic changes and maintain high performance in multi-body environments.
2Productivity
If path planning methods with small time windows (e.g., DWA) are used, then the computation speed is fast, but the planning is limited in time and cannot achieve temporal globalization
Solution Approach 1:
The patent performs preliminary actions by pre-processing environmental data, pre-calculating potential paths, and preparing motion models in advance. This allows the system to extend its effective planning time window while maintaining fast computation speed during actual path planning execution.
Solution Approach 2:
The patent implements dynamic time window adjustment where the planning horizon is adaptively extended or reduced based on the current environmental state, agent velocities, and uncertainty levels. This allows the system to achieve temporal globalization when needed while maintaining fast response when conditions permit.
3Device complexity
If static path planning methods are used, then the planning is simple and fast, but it is only suitable for environments where the agent and task are relatively unchanged
Solution Approach 1:
The patent dynamically changes key parameters such as the planning horizon, agent motion models, and environmental constraints based on the current operational context. This allows the system to maintain relatively simple planning algorithms while adapting to dynamic environments by adjusting parameters rather than fundamentally changing the planning approach.
Data Source
AI summary
The subject matter provides a path planning method, apparatus and electronic device. Wherein, the method comprises: performing environment modeling according to static road network information and dynamic road condition information of a road network so as to obtain an environment model; determining a plurality of candidate paths according to a starting point and an ending point; extracting from the environment model an environmental feature corresponding to each candidate path by a feature extraction network of a path planning model; inputting the environmental feature to a value estimation network of the path planning model so as to obtain an estimated value for each candidate path output by the value estimation network; determining an optimal path among the candidate paths according to the estimated value.