Skip to content

Latest commit

 

History

History
38 lines (30 loc) · 940 Bytes

README.md

File metadata and controls

38 lines (30 loc) · 940 Bytes

Socially-aware-navigation

Overview

Human pose estimation using deep learning method to predict trajectory-based intent from historical human poses using Kalman filter and utilize these predictions for socially aware navigation.

Dataset

MPII Human Pose Dataset

Pipeline

Model Architecture

Results

Preprocessed image with filtered annotations

Model output in simulation

Navigation

Refrences

[1] A plugin for simulation of human pedestrians in ROS Gazebo. https://github.com/robotics-upo/gazebo_sfm_plugin. Accessed: 2022-12-01