A real-time vehicle trajectory prediction method based on the fusion of spatio-temporal awareness graph and multi-scale dilated convolution

Key Points

• The method significantly enhances vehicle trajectory prediction accuracy, potentially benefiting traffic management.
• An average prediction improvement of 20% was noted when using a fusion of spatio-temporal awareness graphs and dilated convolution.
• Analysis involves a real-time prediction method that integrates multiple scales for data fusion, improving response times.
• Highlights may indicate better routing options for automated driving systems, enhancing overall safety and efficiency.