A5026537580- Traffic control and management
- Traffic Prediction and Management Techniques
- Autonomous Vehicle Technology and Safety
- Transportation Planning and Optimization
- Elevator Systems and Control
- Formal Methods in Verification
- Advanced Algorithms and Applications
- Names, Identity, and Discrimination Research
- Traffic and Road Safety
- Face and Expression Recognition
- Transportation and Mobility Innovations
- Neural Networks and Applications
- Earthquake Detection and Analysis
- Islamic Finance and Banking Studies
- Seismology and Earthquake Studies
North China Institute of Science and Technology
2024
Southeast University
2023-2024
This paper proposes a reinforcement learning (RL)-based traffic control strategy integrated with attention mechanism for large-scale adaptive signal (ATSC) system. The proposed RL integrates into multiagent model, namely proximal policy optimization (MAPPO), so as to enable more effective, scalable, and stable in complex ATSC environments. In the RL, decentralized policies are trained using centrally computed critic that shares an while model selects relevant intersections each agent...
To achieve shorter path length and lower repetition rate for robotic complete coverage planning, a complete-coverage path-planning algorithm based on transition probability learning perturbation operator (CCPP-TPLP) is proposed. Firstly, according to the adjacency information between nodes, distance matrix of accessible grid are established, optimal initialization generated by applying greedy strategy matrix. Secondly, population divided into four subgroups, different degrees operations...
This paper proposes an adversarial reinforcement learning (RL)-based traffic control strategy to improve the efficiency of integrated network with expressway and adjacent surface streets. The proposed RL integrates into a multi-agent model, namely advantage actor-critic (MA2C), so as enhance generalization against mismatch between offline-training environment real process. In RL, model is trained maximize throughput, while network, which produces disturbances observed states, based on...
This study introduces CycLight, a novel cycle-level deep reinforcement learning (RL) approach for network-level adaptive traffic signal control (NATSC) systems. Unlike most traditional RL-based controllers that focus on step-by-step decision making, CycLight adopts strategy, optimizing cycle length and splits simultaneously using Parameterized Deep Q-Networks (PDQN) algorithm. effectively reduces the computational burden associated with frequent data communication, meanwhile enhancing...
Diversity control is an important task to alleviate bias amplification and filter bubble problems. The desired degree of diversity may fluctuate based on users' daily moods or business strategies. However, existing methods for controlling often lack flexibility, as decided during training cannot be easily modified inference. We propose \textbf{D3Rec} (\underline{D}isentangled \underline{D}iffusion model \underline{D}iversified \underline{Rec}ommendation), end-to-end method that controls the...
Urban traffic congestion is a critical predicament that plagues modern road networks. To alleviate this issue and enhance efficiency, signal control vehicle routing have proven to be effective measures. In paper, we propose joint optimization approach for in signalized The objective network performance by simultaneously controlling timings route choices using Multi-Agent Deep Reinforcement Learning (MADRL). Signal agents (SAs) are employed establish at intersections, whereas (RAs)...