Hilmi, Faa'iz Haikal (2026) Implementasi Inverse Perspective Mapping Menggunakan ROS 2 Untuk Estimasi Posisi Objek Pada Robot Humanoid Sepak Bola. Other thesis, Institut Teknologi Sepuluh Nopember.
|
Text
5025221219-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy |
Abstract
Estimasi jarak berbasis penglihatan yang akurat merupakan komponen penting dalam sistem persepsi untuk robot humanoid sepak bola. Inverse Perspective Mapping (IPM) menawarkan pendekatan geometris untuk memperkirakan posisi objek dari gambar kamera monocular tanpa sensor jarak tambahan, yang dilarang dalam kompetisi robot humanoid sepak bola seperti RoboCup. Dalam praktiknya, perubahan parameter ekstrinsik kamera secara terus menerus akibat gerakan berjalan dan pergerakan kepala menyebabkan penerapan IPM menjadi tantangan tersendiri. Penelitian ini menjelaskan implementasi IPM untuk estimasi jarak objek pada robot humanoid sepak bola yang dikembangkan oleh tim ICHIRO ITS menggunakan perangkat lunak ROS 2, sebuah kerangka kerja robotika yang banyak digunakan yang memungkinkan pertukaran data secara modular dan real-time antara berbagai subsistem robot, seperti sensor dan modul pemrosesan. Sistem yang diusulkan mengintegrasikan output deteksi objek, parameter intrinsik kamera, dan transformasi kinematika real-time yang diperoleh dari model robot. Pada kondisi statis, sistem mencapai MAE di bawah 8 cm untuk objek di bawah 4 m. Saat robot berjalan (dinamis), kinerja tetap akurat dengan median eror terjaga antara 5,11 cm hingga 14,75 cm. Dibandingkan metode regresi sebelumnya, pendekatan IPM memperluas jangkauan estimasi posisi objek dari 250 cm hingga mencapai 400 cm dengan maksimum eror Euclidean sebesar 4.20 cm, serta menghilangkan ketergantungan pada tracking objek secara aktif.
==========================================================================================================================
Accurate vision-based distance estimation is a critical component of perception systems for humanoid soccer robots. Inverse Perspective Mapping (IPM) provides a geometric approach to estimate object positions from monocular camera images without additional distance sensors, which are prohibited in humanoid robot soccer competitions such as RoboCup. In practice because of camera extrinsic parameters changing continuously caused by walking motion and head movements, IPM is challenging to implement reliably. This study presents the implementation of IPM to estimate object distance on a humanoid soccer robot developed by ICHIRO ITS team using ROS 2 software, a widely used robotics framework that enables modular and real-time data exchange between multiple robot subsystems, including sensors and processing modules. The proposed system integrates object detection outputs, camera intrinsic parameters, and real-time kinematic transformation obtained from the robot model. Under static conditions, the system achieves an MAE below 8 cm for objects within 4 m. During dynamic walking, it maintains robust performance with median errors bounded between 5.11 cm and 14.75 cm. Compared to the legacy regression method, the the IPM approach extends the object position estimation range from 250 cm up to 400 cm with a maximum Euclidean error of only 4.20 cm, while eliminating the operational dependence on active object tracking.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Humanoid robot, IPM, Robot humanoid, ROS 2 |
| Subjects: | T Technology > TJ Mechanical engineering and machinery > TJ211 Robotics. T Technology > TJ Mechanical engineering and machinery > TJ217.2 Robust control |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Faa'iz Haikal Hilmi |
| Date Deposited: | 18 Jun 2026 08:26 |
| Last Modified: | 18 Jun 2026 08:26 |
| URI: | http://repository.its.ac.id/id/eprint/133900 |
Actions (login required)
![]() |
View Item |
