A model training method for predicting obstacle trajectories based on migration scenarios

Abstract

This specification discloses a model training method for predicting obstacle trajectories based on migration scenarios, which relates to the field of unmanned driving. Each training sample, as each candidate training sample, for each candidate training sample, according to the trajectory feature corresponding to the candidate training sample and / or the prediction result of the candidate training sample by the trajectory prediction model, determine the reference corresponding to the candidate training sample value. Select the target training samples according to the reference values ​​corresponding to each candidate training sample, and train the trajectory prediction model according to the target training samples to obtain the trajectory prediction model in the migration scenario, and improve the trajectory prediction model in the migration scenario. While improving the accuracy of trajectory prediction, the data volume of training samples used to train the trajectory prediction model in the migration scenario is reduced.

Description

technical field [0001] This specification relates to the field of unmanned driving, and in particular to a model training method for predicting obstacle trajectories based on migration scenarios. Background technique [0002] With the continuous development of information technology, unmanned driving has initially entered people's lives, and in practical applications, ensuring the safe driving of unmanned equipment is a prerequisite for performing various tasks through unmanned equipment. For the safe driving of unmanned equipment, it is necessary to make unmanned equipment predict the future trajectory of surrounding obstacles, so that unmanned equipment can formulate driving strategies in advance to achieve the purpose of safe driving. [0003] In the existing technology, a large number of obstacle trajectory data can be obtained in history, and a model that can predict the obstacle trajectory can be pre-trained through these trajectory data. There are different effects w...

AI Technical Summary

Problems solved by technology

[0003] In the existing technology, a large number of obstacle trajectory data can be obtained in history, and a model that can predict the obstacle trajectory can be pre-trained through these trajectory data. There are different effects when driving. For example, a model can be trained through a large amount of trajectory data of A, and if the model is applied to the trajectory prediction of the driverless device when driving in B, the predicted trajectory may be inaccurate. Therefore, there is a certain inaccuracy in this method in the prior art, and if the trajectory prediction model is retrained directly through all the trajectory data of B, the required data volume is large and the efficiency is low

Images