Search

LiDAR data widely replaces 2-dimensional data for geographic data representation because of its information complexity. One of the LiDAR data processing tasks is semantic segmentation which has been developed by deep learning models. These algorithms use Euclidean distance representation to express the distance between the points, whereas LiDAR data with random properties are not suitable to use this distance representation. Therefore, this study proposes the non-Euclidean distance representation which is adaptively updated using their covariance values. The proposed method results the accuracy of 75.55%, better than the baseline PointNet of 65.08% and Dynamic Graph CNN of 72.56% with the dataset from the author. This performance improvement is because of multiplication with the inverse covariance value of point cloud data increasing the points similarity to the class.

Three-wheeled mobile robot called “Becak” is one of the alternative transportation that generally used in developing countries. This research proposed a new design of the Becak to become autonomous vehicle one. This paper derived the kinematics model as a fundamental aspect of designing a mobile robot. Three-wheeled kinematic robot model that depending on the back wheel as initial robot velocity and the mobile robot turning angle determined by robot shaft angle. The model is obtained by assuming that a rotation merely depends on the robot shaft movement, as the front wheels both are standard fixed wheel. The rear wheel of the mobile robot also takes a standard fixed wheel. An additional assumption in this model is that there is a slip angle that affected the final pose of the mobile robot. The final pose of the mobile robot then calculated and simulated based on two input variables, i—e., rear-wheel velocity and robot shaft angle. The mobile robot trajectory plots are generated based on the proposed kinematics model.

Three-wheeled bicycle, also known as pedicab, is well known as an eco-friendly transportation in Indonesia. The pedicab is usually operated by utilizing human power to transport objects. It also has high mobility in urban areas which is useful for areas with a heavy traffic. In the future, pedicab which operated by human power can be replaced by autonomous controller. This study aims to develop a model of autonomous pedicab. The kinematics and dynamics model of the pedicab motion were developed and analyzed. Furthermore, autonomous pedicab model is built in the simulated environment. Main components, i.e. sensor, module processing, and engine were added to the conventional pedicab. Initial design of the autonomous pedicab is simulated in Robot Operating System (ROS).

3-dimensional data contains more informative visualization than a 2-dimensional one. LiDAR sensor produces 3D data or point cloud data. There have been many implementations of LiDAR data such as for building detection, urban area modeling, and land cover analysis. This study will analyze land cover because of its substantial benefits. The purpose of this study is to produce semantic segmentation of land cover from LiDAR data by implementing EdgeConv Algorithm from Dynamic Graph Convolutional Neural Network (DGCNN). The dataset in this study is LiDAR data of Kupang, one of the areas in Indonesia. This work achieves the average accuracy of 67.76% for DGCNN better than the state-of-the-art method PointNet (previous method) with 64.97% by implementing the point cloud dataset from LiDAR data of Kupang. This is because the edge convolution could recognize the global shape structure and local neighborhood information so that it could improve the segmentation performance result.

An accurate wind information forecasting plays the significant role for wind power system. However; the intermittent characteristic wind speed in nature over the time and from one location to another makes it hard to estimate the usage factor of wind farms. Therefore, actual long and short duration forecasting of wind speed is necessary for wind power generation system efficiency. In this research, we propose the method to forecast the wind speed data based on weather parameters including, temperature, sea level pressure, dew point, visibility, station pressure, rain intensity, optimum wind speed, maximum temperature, minimum temperature, hail intensity and thunder intensity data. Au parameters were predicted using time series model, then the result of predicted data was implemented to predict the wind speed data. This research implemented radial basis function neural network (RBF NN) to predict the wind speed and the results were compared to univariate time series forecasting and Least Square Support Vector Machine (LS SVM) algorithm. The result experimentally express better forecasting using RBF NN compared to two other models on the measures of MAPE, MSE and correlation coefficient

At this time, the demand of Indonesian mango is in great demand by the society especially for the superior quality mango like Manalagi and Gadung. However, many people are still wrong in distinguishing mango varieties. At this moment, the identification or introduction of mango varieties is done by eye. Some people may be expert in identifying mango varieties based on leaves by eye, but not all mango varieties they can identify. Until now, there are several methods to identify mango varieties, but the accuracy got is less than 80%. In research before, the extraction feature used is either shape or texture feature of the leaf images. In this research, we use Backpropagation Neural Network (BPNN) by using mango leaf shape and texture feature. A Dataset used are 300 images of mango leaves consisting of 150 images of mango leaves of Manalagi varieties and 150 images of leaves Gadung. By using this method, we obtain that the most optimal BPNN model got by using hidden layer = 19, learning rate = 0.9, momentum = 0.9, and epoch = 100 with root mean square error (RMSE) = 0.0018. The accuracy rate that we obtain is 96%