Drone Taxi Route Coordination Using Multi-Agent Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current drone taxi systems face challenges in optimizing routes and maximizing profits due to limitations in existing reinforcement learning technologies, which primarily rely on single-agent approaches and fail to account for cooperative actions among multiple drone taxis.
Innovation Solution
A drone taxi system based on multi-agent reinforcement learning is developed, which includes a control server that uses multi-agent reinforcement learning to optimize routes for multiple drone taxis. This system selects candidate passengers and generates travel route information through a graph neural network, enabling cooperative actions and optimizing routes for maximum profit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If single-agent reinforcement learning is used for route optimization, then the system complexity is low, but the ability to enable cooperative actions among multiple drone taxis is insufficient
Solution Approach 1:
The patent combines multiple single-agent reinforcement learning models into a multi-agent reinforcement learning system where multiple drone taxis operate as independent agents that can cooperate. Each agent maintains its own policy network while sharing the environment and reward structure, enabling cooperative actions through joint state-space exploration and coordinated decision-making.
2Productivity
If multi-agent reinforcement learning is used to optimize routes for multiple drone taxis, then profits are maximized through cooperative actions, but the computational complexity and system requirements increase
Solution Approach 1:
The patent segments the multi-agent system into independent drone taxi agents, each with its own state representation, action space, and neural network policy. This segmentation allows parallel processing of individual agent decisions while maintaining cooperative capabilities through shared environment interaction and coordinated reward optimization.
Solution Approach 2:
The patent changes the learning parameters and architecture of the reinforcement learning system to accommodate multiple agents, including modified state representations that capture multi-agent interactions, adjusted reward functions that reflect collective profitability, and scaled computational resources that optimize the trade-off between cooperative performance and computational complexity.
3Speed
If traditional call taxi systems are used for drone transportation, then the system structure is simple, but the response time and flexibility to environment changes are insufficient
Solution Approach 1:
The patent implements self-service capabilities where each drone taxi agent independently processes passenger requests, autonomously determines optimal routes, and makes real-time decisions without centralized dispatch coordination. This self-service approach enables rapid response to environmental changes while maintaining system simplicity through decentralized autonomous operation.
Data Source
AI summary
Provided are a drone taxi system based on multi-agent reinforcement learning and a drone taxi operation method using the same. The drone taxi system includes a plurality of drone taxies configured to receive call information including departure point information and destination information from passenger terminals present within a certain range and a control server configured to receive call information of passengers from each drone taxi, select a candidate passenger depending on whether a passenger is present, generate travel route information of each drone taxi from drone state information of the plurality of drone taxies through multi-agent reinforcement learning, and transmit the travel route information to the drone taxi.