| Popis: |
Abstract With the acceleration of urbanization in China, the urban population continues to grow, leading to frequent occurrences of crowded public spaces, which in turn trigger traffic congestion and even safety accidents. In order to more effectively control pedestrian flow, enhance the efficiency and safety of public spaces, this experiment conducts in-depth research and improvement on the traditional Dynamic Window Approach (DWA), and applies it to the fine control of pedestrian flow. Specifically, this study comprehensively reviews and analyzes the characteristics of pedestrian traffic flow and the working principles of traditional DWA. Based on this, the shortcomings of traditional DWA in dealing with complex pedestrian flow scenarios are identified, and targeted improvement solutions are proposed. The core of this improvement scheme lies in the introduction of a new evaluation function, enabling DWA to more accurately balance various factors in the decision-making process, including pedestrian movement speed, direction, and spatial distribution. Subsequently, the improved DWA is validated through simulation experiments. The experimental scenario is set in an area of 18 m*18 m, and compared with traditional DWA, the improved DWA shows significant advantages in trajectory length and travel time. Specifically, the trajectory length of the traditional DWA robot is 19.4 m, with a required time of 34.8 s, while the trajectory length of the improved DWA robot is shortened to 18.7 m, and the time is reduced to 18.6 s. This result fully demonstrates the effectiveness of the improved DWA in optimizing pedestrian flow control. The improved DWA proposed in this study not only has strong scientific validity but also demonstrates high efficiency in practical applications. This study has important reference value for improving the safety of urban public spaces and improving pedestrian traffic flow conditions, and provides new ideas for the further development of pedestrian flow control technology in the future. |
