Crowd trajectory prediction method and system based on deep convolutional long-short memory network

Abstract

The invention discloses a crowd trajectory prediction method and system based on a deep convolutional long and short memory network. The method comprises the steps of firstly determining a pedestriantrajectory data set; secondly, determining a training set and a test set according to the pedestrian trajectory data set; constructing a Conv-LSTM model of the deep convolution long-short term memorynetwork again; determining parameters of a deep Conv-LSTM model according to a training set and the test set; and finally, performing crowd trajectory prediction according to the deep Conv-LSTM modelwith the parameters. According to the method, a Conv-LSTM (Convolutional Long Short Term Memory) model of the deep convolution long-short term memory network is constructed; a deep Conv-LSTM model isused to carry out crowd trajectory prediction; on one hand, the complexity of the network and the prediction precision of the crowd trajectory are improved, the time and space memory of the network for the historical trajectory of the crowd is deepened, and on the other hand, the regression capability of the network is further improved by adopting the context semantic space information output by the convolutional network for the Conv-LSTM layer.

Description

technical field [0001] The invention relates to the technical field of traffic simulation and robot path planning, in particular to a crowd trajectory prediction method based on a deep convolutional long-short memory network. Background technique [0002] With the increasing urban population density around the world, the safe evacuation of pedestrians in emergency situations has become an important issue, and the prediction of pedestrian trajectories has attracted the attention of society and the government. At the same time, in the field of intelligent robots, although the existing robots have made breakthroughs in life communication, motion stability, etc., the path planning of robots is still a big problem. In essence, the prediction of pedestrian trajectories provides new ideas for robot path planning to a large extent. Traditional path planning is considered to be that the robot searches for the optimal non-collision path from the start point to the end point in a dyna...

AI Technical Summary

Problems solved by technology

Although these studies have combined the laws of pedestrian movement, due to the unreasonable network design and only relying on the data of a single scene training, in the test scene, some pedestrians will cross obstacles or have abnormal trajectories.

Especially in emergency evacuation scenarios, their neural network trajectory prediction ability is not high

Images