Anhar Risnumawan

Category: Humanoid Robotics

EROS is a humanoid robot that actively competing in Kontes Robot Sepak Bola Indonesia Humanoid and Robocup Humanoid Soccer League. In these contests, robot must be able to play soccer like a human such as looking for the ball, coordinating with another robot, taking the role, etc. Localization plays a crucial role in perfecting these skills because localization helps the robot know its position and orientation in the field. In this final project, we developed a localization system using Particle Filter method. We used robot’s IMU and kinematic data to update particle’s position and orientation. We used OpenCV DNN Module to detect object (landmark) from image that captured by single camera. We used pinhole camera model to measure the distance of robot towards landmark. The results shows that pinhole camera model able to measure the distance with 402mm average error and 6.59% average percentage error. Particle Filter algorithm with 1.000 particles configuration can achieve position estimation with 429mm and 282mm average error at x and y axis, 11.46ms average estimation process, and 0.2 degrees average error in orientation.

Category: Mobile Robotics, Control System

Framework sistem pemetaan menggunakan metode SLAM GMapping dan sistem navigasi menggunakan metode Adaptive Monte Carlo Localization dan Dynamic Window Approach berbasis ROS yang dapat secara langsung diimplementasikan pada berbagai macam autonomous mobile robot (AMR) solusi physical distance guna mempercepat pengembangan robot untuk perangi COVID-19.

Category: Humanoid Robotics

Humanoid robots are robots that are formed based on the human body and are able to interact with equipment and environments made for humans. In designing a humanoid robot, the balance of the robot is the basic thing to do various activities. Establishment of control on the ROS platform with the target to obtain stable running. The system construction design is carried out to connect the ROS platform with the robot. The design of software construction starting from virtual robot visualization can be done by making a simulator that contains dynamic parameters. The humanoid robot movement can be tried many times until the robot movement is robust with the simulation. A simulator containing dynamic z parameters can be made by applying the URDF Model to the Gazebo simulator and supported by the ROS framework. Feedback in the form of position and torque is needed to determine the difference between simulation and reality. The feedback is also to find out the truth of the virtual robot visualization. Simulations can be carried out without cost, risk, and most importantly imitating the real soccer robot environment and being tested in real conditions.

Category: Embedded System, Industrial Automation

STAMP is a stator condition monitoring system on an induction motor where data comes from previous research and later detection results will be shown in real time using the Internet of Things (IoT). The realization of IoT is poured in the form of a web service. And there is a MyRio LED on this system that is used as a Warning System for normal or broken conditions.

Category: Humanoid Robotics, Computer Vision

Humanoid Robot Soccer merupakan robot yang menyerupai manusia dan mampu melakukan interaksi dengan lingkungan di sekitarnya termasuk bermain sepak bola dengan aturan-aturan yang telah dibuat. Dalam perancangan robot ini, sistem kendali yang mampu bertahan dengan baik dalam kondisi lapangan sangat dibutuhkan. Penelitian ini berfokus pada sistem citra robot untuk mengenali lingkungan di sekitarnya. Dari pengenalan objek tersebut robot diharapkan juga mampu mengetahui posisi global dari objek-objek tersebut maupun posisi robotnya sendiri. Dalam penerapannya, selain data citra yang didapat dari kamera, robot memerlukan data kinematik posisi robot untuk dapat mengetahui dengan akurat posisi robot di lapangan. Robot harus bisa berdiri tegak agar pendeteksian dan estimasi posisi bisa dilakukan sehingga dalam penelitian ini, robot diharuskan bisa berdiri. Gerakan robot dilakukan secara virtual menggunakan gazebo simulator yang didukung framework ROS sebelum menerapkan secara langsung di lapangan karena simulasi dapat dilakukan tanpa biaya dan resiko kerusakan. Tercapainya posisi robot yang sama dengan kondisi virtual menjadi syarat sebelum dilakukan pendeteksian dan estimasi posisi. Objek yang dideteksi adalah gawang, garis lapangan, bola dan halangan. Persentase keberhasilan pendeteksian lapangan dan halangan adalah 100%, untuk tiang gawang kiri dan tiang gawang kanan adalah 75%, sedangkan untuk bola dan garis lapangan secara berturut adalah 83% dan 41%. Semua objek yang berhasil dideteksi dan diestimasikan posisinya.

Category: Humanoid Robotics

Humanoid robot soccer is a robot whose overall appearance is formed based on the human body that has the purpose of playing football like humans. Therefore, humanoid robot soccer has sensors like the eye that is the camera and other sensors approaching the human body. In designing humanoid robot soccer, robot design becomes fundamental so that the robot has a shape and has a movement like humans. Taken as an example from the Indonesian Football Robot Contest (KRSBI) humanoid division, robots must have the ability to be able to walk, run, and kick the ball dynamically while maintaining balance. In this final project a humanoid robot soccer design with 87 cm height has carried out. Furthermore, kinematic static and dynamic analysis has performed on humanoid robots. This kinematic analysis is used to control robot movements by actuators to match the moves as desired, while dynamic analysis is used to maintain the robot's balance when moving / walking. From the above analysis, the robot balances when there is a disturbance of 30 degrees when walking in place and produces an accurate movement of each servo following the trajectory reference.

Category: Computer Vision, Medical

Corona Virus Disease 19 or better known as Covid-19 is one of the health problems that need our attention today. This's because Covid-19 can be transmitted very easily to humans and can cause a considerable amount of deaths. One way to find out someone is indicated by Covid-19 is to measure body temperature. The technology that has been used tends to be manual for example by checking the temperature with an infrared thermometer gun at the entrances, or temperature screening at airports in some countries with high-resolution thermal cameras that fix and rigid certain places with updates and maintenance that is not easy. Therefore, a portable temperature detection tool is very important to develop. Not only temperature as a detection parameter, but also detects breathing intensity and heart rate. This device is made portable and connected to a local network so that it is easy to install and accessible. The results obtained from the work so far are that the portable device system can work according to its function and can detect heart rate and respiratory rate with an RMSE value of 6.83 for heart rate and 8.47 for respiratory rate based on comparisons with related medical devices and manual calculations.

Category: Humanoid Robotics

Humanoid robot soccer is a type of robot that has a body shape like a human and a special program for playing soccer. Surabaya State Electronics Polytechnic (PENS) has a special research team in this field, namely the EROS team. The EROS team actively competes in the Indonesian Humanoid Soccer Robot Contest and the Robocup Humanoid Soccer League. In these contests, robot games to be able to play soccer like humans. The balance factor plays a crucial role in helping humanoid robots to be stable. The EROS humanoid robot already uses a balance control system when the robot is running, but the balance control system is still not adaptive. The tilt angle of the robot's body is set manually based on the movement carried out by the robot. So it takes time for the process of tuning the large value of the tilt angle of the robot body. In this final project, an adaptive balance control system was designed. The robot's gyroscope data is used to move the robot's legs when it detects changes in the tilt of the robot's body. The leg movements are carried out to maintain the stability of the robot. The adaptive balance control system can help correct the tilt of the humanoid robot's body to keep it centered.

Category: Computer Vision, Image Processing

Prehistorical sites are one of the media for people to study about the vast prehistorical events. Indonesia is a country that has so many prehistorical sites. In addition, one of the Indonesian sites, Sangiran, is one of the most important sites according to UNESCO. These sites keep so many historical objects such as fossils or artifacts. These objects are so important for humanity because they contain historical stories. Hence, historical objects should be kept and well maintained. But at the other side, visitors of sites would hardly access the details of historical objects. 3D Reconstruction is a correct solution for the problem. The usage of Structure from Motion Technique (SfM) could make 3D reconstruction of historical objects in point clouds format. SfM Technique could be applied on photos that captured by camera with photogrammetry technique, the camera captures the image of objects from many different angles and the background of the images would be blurred with 75 pixel radius blur before processed with SfM algorithm to obtain the 3D reconstruction. The dataset could be stored in the museum's online page database or printed with a 3D printer so people could easily learn the details of historical objects.

Category: Mobile Robotics, Artificial Intelligence, Control System

ERSOW (EEPIS Robot Soccer On Wheeled) merupakan tim robot sepak bola beroda Politeknik Elektronika Negeri Surabaya (PENS) yang berlaga dalam Kontes Robot Sepak Bola Indonesia Beroda. Penelitian dalam bidang robot sepak bola beroda banyak dilakukan pada kecerdasan robot seperti bagaimana robot mendeteksi dan mencari bola, menggiring bola, mengumpan bola, menghindari lawan, dan komunikasi robot dalam tim. Dalam rancangan tema Tugas Akhir ini difokuskan pada bagaimana robot melakukan passing, yaitu kemampuan robot dalam hal mengumpan dan menerima bola. Robot mengetahui posisi robot teman dengan deteksi objek cover dan komunikasi antar kedua robot menjadi faktor penting keberhasilan passing. Robot pengumpan menyesuaikan dengan posisi robot penerima dan selanjutnya melakukan umpan ke robot penerima. Robot penerima akan menyesuaikan orientasinya terhadap robot pengumpan berdasarkan bola agar bola dapat ditangkap dengan baik. Hasil pengimplementasian algoritma passing telah dilakukan pada latihan dan lomba Kontes Robot Sepak Bola Beroda Indonesia pada tahun 2021. Persentase keberhasilan kedua robot dalam menjalankan algoritma passing sebanyak 423 kali adalah 98.83%.

Category: Agriculture

Agricultural Technology

Category: Mobile Robotics, Computer Vision

The 3D photogrammetry mobile robot was designed to be an autonomous image-capturing system to provide the input dataset for the 3D reconstruction process. The research aims to develop a mobile robotic platform to move around a certain object, which is in this case a batik display, and get pictures from various angles with tracking and adjustment features using a Node-RED-based graphical user interface. This mobile robotic platform is developed to ease the users to get numerous pictures for the photogrammetry process without manually moving the camera from one place to another. This project comprises of a differential drive mobile robot with a turnable camera holder on it and a master computer unit as a main controller and server to be accessible by the user. The system was designed to satisfy the requirements of orientation stability and less vibrating behavior to get a high-definition picture provided with a blurred detection system on a graphical user interface to delete the blurry image from the collection of pictures. The result of obtained data from the experiment including 0.006% error average position rate, a threshold more than 10 of Laplacian kernel on blurred detection to get a more accurate classification result, and the root means square error (RMSE) on the 3D reconstruction process close to zero.

Category: Agriculture

Penelitian ini merancang dan membangun sebuah sistem robot yang inovatif untuk mengatasi permasalahan kurangnya nutrisi pada tanaman serta mencegah pertumbuhan gulma yang merugikan dengan cara yang ramah lingkungan dan tanpa bahan kimia berbahaya. Sistem ini menggabungkan teknologi Artificial Intelligence (AI) dengan arsitektur YOLOv6 untuk mendeteksi objek, termasuk dua jenis tanaman pokok (pokcoy usia 1 minggu dan 4 minggu) serta gulma berjenis rumput teki. Sistem robot ini menggunakan 3 axis (X, Y, Z) yang dikendalikan oleh sistem benam dan dilengkapi dengan kamera untuk mendeteksi keberadaan gulma pada papan tanam. Setelah terdeteksi, sistem secara otomatis menentukan koordinat gulma dan mengarahkan alat pencabut dengan akurasi 98,6% dalam penanganan gulma dan hanya 1,4% tingkat error pada akurasi koordinat pencabutan gulma. Implementasi teknologi AI dengan arsitektur YOLOv6 memberikan hasil deteksi yang sangat baik dengan akurasi sekitar 90% untuk jenis gulma rumput teki serta mampu membedakan jenis tanaman dengan gulma secara efisien. Dengan aplikasi yang memungkinkan monitoring dan pengendalian jarak jauh, sistem ini memberikan solusi yang inovatif untuk memudahkan petani rumahan dalam merawat tanaman.

Category: Artificial Intelligence, Computer Vision, Machine Learning

Visually Impaired is a condition with a lack of eye function experienced by someone that makes the visually impaired unable to see or blind. The factors that cause a person to become blind are divided into two, and the First is Prenatal and Postnatal. According to the Indonesian Ministry of Health estimation, the number of visually impaired in Indonesia is 1.5% of the total population. Visually Impaired has a lack of function in the eyes that makes a person unable to do certain jobs normally or even unable to do some work. In comparison, about 80% of the information we receive comes from the eyes. This matter makes the visually impaired have an attitude of dependence because they cannot be independent in their activities and are not skilled in orientation and mobility. A stick is a commonly used tool to improve orientation and mobility skills in Indonesia. The stick functions as an assistant for the visually impaired and allows them to travel independently and safely. However, using sticks as a tool for the visually impaired makes detecting obstacles around them difficult and determining the final location so that relatives or families can find the visually impaired when users feel lost. Smart Stick is a tool for the visually impaired to detect obstacles around and determine the last location that can help the visually impaired contact their family when they feel lost. Smart sticks can improve the orientation and mobility of the visually impaired through object detection using Convolutional Neural Network with an mAP of 99.42% of the five predetermined classes, namely car, pedestrian, pothole, motorcycle, and cycle, with 6000 iterations of training data and sending the last location to the family in the form of a link google maps with 6.14 m distance tolerance via Short Message Service (SMS) which visually impaired families can access.