Trajectory prediction method and system for vulnerable road users based on motion query enhancement

By expanding the static query and optimizing the dynamic query in the trajectory prediction method for vulnerable road users, the problems of insufficient coverage of static queries and weak connection of dynamic queries are solved, and more accurate future trajectory prediction is achieved.

CN120216607BActive Publication Date: 2026-03-17XI AN JIAOTONG UNIV
0 Cites 0 Cited by

Patent Information

Application Number
CN202510297766.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-03-13
Publication Date
2026-03-17
Estimated Expiration
2045-03-13

AI Technical Summary

Technical Problem

In existing methods for predicting the trajectories of vulnerable road users, static queries cannot effectively cover the boundary samples of motion patterns, and dynamic queries are not tightly connected in the cascaded decoder, resulting in inaccurate prediction results.

Method used

The current position and orientation of motion samples are aligned by normalization, and static queries are generated by clustering. The observation sequence is added to the static queries to expand their effective range. Cross-connection dynamic queries are used in the cascaded Transformer decoding layer, and trajectory and map features are aggregated by the cross-attention module to optimize the dynamic queries. Gaussian mixture models are used to predict future trajectories, and the final result is selected by an integration strategy of similar motion patterns.

Benefits of technology

It enhances the coverage of static queries, optimizes the effectiveness of dynamic queries, and improves the accuracy of predicting the future trajectories of vulnerable road users.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application provides a weak road user trajectory prediction method and system based on motion query enhancement, which comprises the following steps: aligning the current position and direction of each motion sample through a normalization process, clustering the end points of the normalized motion samples, taking the obtained cluster centers as generated static queries, using extended static queries (DSQ) to extend the coverage of the static queries to provide more diverse motion patterns; combining the extraction and fusion of trajectory features and map features to obtain map features and enhanced trajectory features, which are used to obtain the update query features of each layer subsequently, using cross-connection dynamic queries (BDQ) to establish a connection between the dynamic queries output by the shallow and deep decoders, coordinating the query optimization between the decoding layers; and integrating the prediction results of multiple homogeneous models to obtain the final prediction trajectory. The application can more effectively extract features that can reflect the motion trend of weak road users, making the prediction results more accurate.
Need to check novelty before this filing date? Find Prior Art