Highway automatic driving commercial vehicle trajectory prediction and decision method and system based on complex network theory

By employing a trajectory prediction and decision-making method based on complex network theory, the self-collision problem in multimodal prediction in autonomous driving is solved, enabling safer and more diverse trajectory prediction and improving the decision-making accuracy and interpretability of autonomous driving.

CN116050245BActive Publication Date: 2026-06-09JIANGSU UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
JIANGSU UNIV
Filing Date
2022-12-05
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing autonomous driving trajectory prediction methods suffer from self-collision risk in multimodal prediction, do not fully consider scene consistency and obstacle interaction, and lack interpretability.

Method used

A trajectory prediction and decision-making method based on complex network theory is adopted. By constructing a dynamic complex network model, a complex network encoder and a memory network are used to predict the target point of the agent. By combining the reachability of lane nodes and the criticality of agent nodes, a joint distribution of reference trajectories is generated. A trajectory decoder is used to generate predicted trajectories that meet the constraints. Finally, the decision-making model selects the optimal trajectory.

Benefits of technology

It improves the scenario consistency and interpretability of trajectory prediction, avoids the risk of collision between the vehicle and other intelligent agents, conforms to human driving habits, and provides safer and more diverse trajectory prediction results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116050245B_ABST
    Figure CN116050245B_ABST
Patent Text Reader

Abstract

The application discloses a highway automatic driving commercial vehicle trajectory prediction and decision method and system based on complex network theory, which utilizes a complex network encoder in a prediction model to encode a high-definition map and an agent context into node coding of a directed weighted complex network to obtain the accessibility of each node.A target point predictor predicts target points of each agent according to historical trajectory information and local road features of the agent, and obtains a joint distribution of reference trajectories in combination with lane node accessibility and agent node criticality.A trajectory decoder inputs historical trajectories of the agent and kinematics or dynamics constraints to decode and generate a predicted trajectory and optimize the predicted trajectory under the condition of the reference trajectory.A decision model inputs a navigation route provided by an upstream global path planning layer, judges whether there is a lane changing requirement, and determines a priority of a lane.Combining a joint distribution probability of the reference trajectory, a complex network risk value and the lane priority, a cost function is designed to select an optimal trajectory as a decision result.
Need to check novelty before this filing date? Find Prior Art