Please use this identifier to cite or link to this item:
Title: Prediction of pedestrian trajectory in a crowded environment using deep learning
Authors: Xiong, Xincheng
Keywords: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
Issue Date: 2019
Abstract: Deep learning is the current state-of-the-art technique for most machine learning and data analytics tasks. It has been applied to all aspects of human life. Traditional methods are not suitable for solving pedestrian trajectory prediction due to the diversity of its scenes and the uncertainty of the original trajectory. However, problems that cannot be solved by traditional methods are not unsolvable for deep learning. The pedestrian trajectory prediction problem is essentially a prediction problem. Due to the great success of the RNN architecture in sequence prediction, the RNN architecture is our best choice for solving problems. In our project, in order to solve the problem that the RNN architecture does not have great accuracy when the input is long sequence, we use the LSTM model [28] for experiments. The LSTM model typically produces higher accuracy when processing long sequence inputs [7]. Based on LSTM, we built the Encoder-Decoder model which can output predictive coordinate sequences through a serial human coordinate sequence input. By continually trying, we have found model parameters ranges that make the model own better accuracy. Predicting pedestrian trajectory from its past trajectory in a dynamic environment is a very creative idea. The results show that our model already had considerable accuracy in predicting pedestrian trajectories. Although the accuracy is not high in some scenarios, this is a great achievement in such problems.
Schools: School of Electrical and Electronic Engineering 
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
FYP report.pdf
  Restricted Access
4.51 MBAdobe PDFView/Open

Page view(s)

Updated on Sep 25, 2023


Updated on Sep 25, 2023

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.