The research develops an autonomous package delivery car robot system using the A* algorithm on 2D Point Cloud data from LIDAR and object detection with YOLO. The pathfinding system using the A* algorithm successfully navigates the robot with a success rate of 100%, an average processing time of 1.6 ms, and an average cost of 8.09. The robot is designed with specifications of operating temperature of-40 degrees C to +85 degrees C, operating time of 180 minutes, and maximum payload capacity of 5 kg. Object detection testing using YOLOv4-tiny on 2D Point Cloud resulted in an Average Precision (AP) of 100% for cars and humans, 92.74% for motorcycles, with a detection sensitivity of 68.3%. The distance prediction system has 97% accuracy, with an average prediction difference of 0.092 meters. This research shows that the application of the A* algorithm to 2D Point Cloud data improves the efficiency and accuracy of robot movement, and makes a significant contribution to the development of autonomous vehicles for logistics applications.
Predicting the unemployment rate with precision is essential for policymakers, economists, and business leaders, as it enables effective decision-making in economic planning and resource distribution. This research conducts a comparative evaluation of two prominent machine learning methods-Deep Learning and XGBoost-for forecasting Indonesia's Open Unemployment Rate (OUR). Drawing on data from Indonesia's Central Bureau of Statistics, the study builds and assesses two predictive models: one based on Deep Learning and the other on XGBoost. Model performance is measured using key metrics such as Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Success Percentage. Findings reveal that XGBoost surpasses Deep Learning in accuracy for predicting Indonesia's OUR. Although the Deep Learning model excels at identifying intricate patterns, XGBoost provides superior interpretability throughout training and testing. This analysis underscores the advantages and drawbacks of each method in unemployment forecasting, offering actionable insights for subsequent research and real-world economic prediction tools.
Driven by the shortcomings of traditional Proportional Integral Derivative (PID) controllers, particularly their limited responsiveness to changes in system parameters, suboptimal performance in nonlinear systems, and challenges in managing disturbances and noise, this study explores an alternative approach. Long Short-Term Memory (LSTM) controllers, designed to mimic PID controller behavior, are anticipated to effectively learn and predict nonlinear system dynamics, exhibit robustness against disturbances and noise, and automatically adapt their parameters. This paper proposes emulating a PID controller using an LSTM network, with the Internet-Based Temperature Control Lab (iTCLab) Kit as the test platform. The results demonstrate successful modeling and emulation of an adaptive PID controller using LSTM on the iTCLab, evidenced by satisfactory control performance, characterized by a relatively slow rise time, minimal overshoot, and low steady-state error.
The salt industry faces declining sales due to the lack of automated processing machines, affecting efficiency and market readiness. This study develops a salt refining and stirring machine using SUS 316L, known for its durability and hygienic properties. The machine, with a maximum rolling speed of 3800 rpm, refines salt to a smoothness level of 50-100%, depending on the roller gap density. Testing shows that refining 1 kg of salt with a 1 mm roller density takes 6.01 minutes (5mm gap), 1.13 minutes (10mm gap), 0.1 minutes (15mm gap), and 0.07 minutes (20mm gap). For a 1.5 mm roller density, times reduce to 5.2, 0.95, 0.08, and 0.04 minutes, respectively. The integration of IoTbased monitoring with ThingSpeak enhances efficiency through real-time voltage and current tracking, predictive maintenance, and remote monitoring, reducing downtime and optimizing performance.
The rapid advancement of 3D reconstruction technology has created significant opportunities across various fields, yet its complexity remains a barrier to widespread adoption, particularly for novice users. This research evaluates and compares traditional and modern 3D reconstruction methods to assess their effectiveness in terms of reconstruction quality and processing efficiency. Classical approaches such as COLMAP and VisualSfM are analyzed alongside the neural network-based Instant-NGP to highlight their respective strengths and limitations. A comparative study is conducted based on both objective metrics and subjective human perception, ensuring a comprehensive evaluation of their performance. Additionally, user feedback is collected to assess ease of use and accessibility, providing insights into potential improvements for broader adoption. The findings indicate that modern deep learning-based approaches offer significant advantages in speed and flexibility, while classical methods retain strengths in accuracy and consistency. To facilitate access to 3D reconstruction frameworks and ensure a more reliable user evaluation, we also incorporate a web-based interface. This eliminates the need for users to manually collect data and execute reconstruction steps, allowing them to focus solely on evaluating the final 3D reconstruction results, thereby enhancing the validity of the user survey results.
