People detection on 2D laser range finder data using deep learning and machine learning - Study on integrating monocular camera and 2D LRF for efficient real-world person detection

Aug 5, 2026

Upload by Liam

Machine learning methods are used for people detection with 2D Laser Range Finders (LRFs), strengthening mobile-robot performance in varied environments. The study integrates a monocular camera with a 2D LRF to improve detection and tracking accuracy and efficiency, while deep learning models such as CenterNet support automatic dataset labeling from combined image and range data. Two experiments are conducted: controlled simulations and real, office-like tests. Results show strong human-leg discrimination, with XGBoost achieving the best accuracy, precision, recall, and F1-score, balancing precision and computational efficiency for potential real-time use.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Mobile robot people detection background

Laser-based approaches and motivation

Organization of the work

Description of the proposal

Simulated testing with CoppeliaSim

Real-world validation setup and sensor integration

People detection on 2D laser range ✜nder data using deep learning and machine learning⋆ José Abrego-González, Eugenio Aguirre, and Miguel García-Silvente Department of Computer Science and A.I. (DECSAI) . Andalusian Research Institute in Data Science and Computational Intelligence (DaSCI) . CITIC-UGR. , University of Granada (UGR), 18071 Granada, Spain [jabrego@correo.ugr.es](jabrego@correo.ugr.es), [eaguirre@decsai.ugr.es](eaguirre@decsai.ugr.es), [m.garcia-silvente@decsai.ugr.es](m.garcia-silvente@decsai.ugr.es) Abstract. This work presents a machine learning based study on people detection using 2D Laser Range Finders (LRFs) combined with deep learning methodologies, aimed at enhancing mobile robot capabilities in various environmental conditions. The study introduces a novel integration of a monocular camera with an LRF on a mobile robot to improve the accuracy and e✣ciency of detecting and tracking people. By employing deep learning models such as CenterNet, the system leverages both image and 2D range data to facilitate automatic labeling of datasets, crucial for training robust classi✜cation algorithms. In order to achieve the best classi✜er, two experimental studies are introduced in this work. The former is carried out in a simulated environment and the latter in real-world, o✣ce-like environments. In simulations, various machine learning models are trained and evaluated, showing signi✜cant results in distinguishing human legs from other objects. The transition to realworld testing underscores the challenges and adaptations necessary to achieve high accuracy and reliability in dynamic settings. The XGBoost model emerged as the most e✛ective classi✜er in our study, achieving the highest scores in accuracy, precision, recall, and F1-score, outperforming other methods across these key metrics. This work aims to advance the ✜eld of 2D LRF based people detection and also proposes a solution for real-time applications, balancing precision and computational e✣ciency. Experimental results from both simulated and real-world environments demonstrate the system✬s e✛ectiveness. Keywords: People detection ➲ Deep learning ➲ Machine learning ➲ 2D LRF 1 Introduction Mobile robots rely on detecting and tracking people for applications like HumanRobot Interaction (HRI), navigating crowded spaces, and safety in shared environments [20] . Various computer vision techniques using Monocular, Stereo, and ⋆ This work was made possible thanks to the support of Senacyt Panamá(Scholarship No. 270-2022-164) and Grant PID2022-138453OB-I00 funded by MCIN/AEI/10 . 13039/501100011033 and by ✏ERDF A way of making Europe✑ . RGB-Depth cameras, including deep learning methods like YOLO, have provene✛ective for these tasks [15] . Despite the advantages of vision systems, 2D Laser Range Finders (LRFs) are favored in social and service robots for their reliability and wide ✜eld of view, overcoming the limitations of vision sensors in adverse conditions [4] . Detecting people using 2D laser technologies can be achieved through various approaches, as outlined in the survey conducted by M. Sharif [22] . Some techniques directly process laser measurements individually as inputs for supervised machine learning algorithms. Conversely, alternative methods ✜rst cluster these measurements and then derive features to characterize these clusters. In this work, we will adopt the latter approach and propose a set of innovative features compared to those described in the specialized literature. Emerging studies utilize deep learning for enhanced detection from sensor data, o✛ering signi✜cant improvements in reliability [21] . Given the absence of automatic labeling tools for 2D laser data, this work explores the potential of deep learning to automate the labeling of such datasets and the use of machine learning approaches to generate e✣cient leg detectors, aiming to enhance e✣-ciency and accuracy in diverse applications [14] . This work is or

FAQs

How does the proposed system combine monocular camera data with 2D LRF data?

A monocular camera is integrated with a 2D LRF mounted on a mobile robot. The system captures both image and 2D range data together with odometry and velocity for later analysis.

What role does deep learning play in dataset preparation and detection?

Deep learning models such as CenterNet use the combined image and 2D range data to enable automatic labeling of the dataset, which is then used to train robust classifiers.

Which classifier performed best and on what metrics?

XGBoost provided the highest scores across accuracy, precision, recall, and F1-score, outperforming other methods in the study.

Aug 5, 2026

Upload by Liam