Paper #5

Research on Unmanned Path Planning of Intelligent Vehicle Based on Swarm Intelligence Algorithm

Tao Yang, Dandan Song

Abstract: Unmanned vehicles represent a significant advancement in automotive technology, with their development hinging on sophisticated perception, decision-making, and control capabilities. However, existing path planning methods for driverless cars face challenges in complex road environments due to their susceptibility to environmental factors. This paper aims to address this issue by first providing an overview of trajectory planning algorithms for driverless cars. Subsequently, a novel global path planning approach is proposed, leveraging an improved A* algorithm and a predictive model of travel time. The proposed method enhances path planning accuracy by integrating the A* algorithm with predictive capabilities regarding travel time. By considering not only the shortest path but also the anticipated time required to traverse it, the model can account for dynamic factors such as traffic congestion and road conditions. This predictive aspect adds a layer of adaptability to the path planning process, enabling intelligent vehicles to make informed decisions in real-time. Simulation results demonstrate the efficacy of the proposed model in accurately planning trajectories for intelligent vehicles. The research results indicate that the prediction results of the bidirectional LSTM network are highly consistent with the actual values, demonstrating good predictive ability. From the perspective of prediction error, the MAE (Mean Absolute Error) of the bidirectional LSTM model is 7.3165, which is superior to the other three models. Especially compared with unoptimized BPNN, bidirectional LSTM reduced MAE, MAPE, and RMSE by 32%, 38%, and 3%, respectively, which fully demonstrates the advantages of bidirectional LSTM in processing time series data. It can accurately predict the inflow of road segments in real time and calculate the travel time of a future road segment.

Keywords: Autonomous driving; Bidirectional LSTM network; Path planning; Predicting inflow volume; Swarm intelligence algorithm; Vehicle trajectory.