A Trajectory Prediction Method for Unmanned Vehicles Based on Local Attention Mechanism

A trajectory prediction, unmanned vehicle technology, applied in prediction, neural learning methods, computer parts and other directions, can solve the problem of insufficient excavation of the interaction between surrounding vehicles and unmanned workshops, rough historical trajectory information, low trajectory prediction accuracy, etc. question

Active Publication Date: 2022-05-31
HUBEI UNIV OF AUTOMOTIVE TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to solve the problem that in the current trajectory prediction field, most methods only roughly consider the historical trajectory information of surrounding vehicles at all times, and do not fully explore the interaction between surrounding vehicles and unmanned workshops. The phenomenon of low trajectory prediction accuracy

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Trajectory Prediction Method for Unmanned Vehicles Based on Local Attention Mechanism
  • A Trajectory Prediction Method for Unmanned Vehicles Based on Local Attention Mechanism
  • A Trajectory Prediction Method for Unmanned Vehicles Based on Local Attention Mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0042]

[0043] Among them, the encoder of the LSTM model is responsible for encoding the trajectory information of each vehicle into the hidden state.

[0048] Input the historical trajectories of all vehicles around the unmanned vehicle into the encoder of the LSTM model, based on the unmanned vehicle cycle at time t.

[0049] The encoded hidden state vectors of all surrounding vehicles at the current time t are changed linearly.

[0050]

[0051] Among them, , is the weight matrix learned through backpropagation.

[0058]

[0061] Compare the hidden state vector of the i-th vehicle around the window at time t, one by one with the hidden state vector of the unmanned vehicle at time t.

[0062]

[0065]

[0067]

[0073]

[0075]

[0079].

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a trajectory prediction method for an unmanned vehicle based on a local attention mechanism, which takes the historical trajectory of vehicles around the unmanned vehicle as input, and fully considers the influence of the interaction between the unmanned vehicle and adjacent vehicles on the future trajectory of the unmanned vehicle; According to the geometric structure of the road and the geometric shape of the vehicle, the spatial interaction between the unmanned vehicle and the adjacent vehicle is constructed, and then some vehicles with a high correlation with the future trajectory of the unmanned vehicle are estimated through the local attention mechanism, and the relationship between these vehicles and the unmanned workshop is calculated. The correlation degree, using the weighted sum of the correlation degree to construct the time interaction; integrating the time and space interaction between the unmanned vehicle and the surrounding vehicles at the current moment, input the comprehensive interaction features and then connect to the decoder of the fully connected layer to obtain the unmanned vehicle for a period of time in the future The trajectory distribution and trajectory coordinates of people and vehicles; during training, the negative logarithmic likelihood loss function is used to calculate the loss, and the parameters are updated through loss backpropagation. The trained model predicts the trajectory of the unmanned vehicle for a period of time in the future, and assists the completion of subsequent decision-making planning.

Description

A Local Attention Mechanism-Based Trajectory Prediction Method for Unmanned Vehicles technical field The invention belongs to the field of intelligent driving, and specifically refers to a kind of unmanned vehicle trajectory prediction based on local attention mechanism method. Background technique In recent years, with the advent of intelligent driving upsurge, the application of artificial intelligence technology in automobiles is increasing day by day, especially for This is especially true for vehicles that are committed to becoming purely driverless. And as a trajectory prediction that predicts the car's position in the next second Technology is the basis for the realization of unmanned driving. Only by correctly predicting the future position of the vehicle can the subsequent actions not affect the vehicle. If it is predicted that the vehicle is about to leave the current lane, it can be predicted in advance whether this action will cause danger. If it will ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06Q10/04G06K9/62G06N3/04G06N3/08
CPCG06Q10/04G06N3/084G06N3/044G06N3/045G06F18/214
Inventor 杨正才石川周奎姚胜华张友兵尹长城冯樱刘成武
Owner HUBEI UNIV OF AUTOMOTIVE TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products