In this tutorial, I explain the concept, probabilistic sensor fusion model and the sensor model used in Octomap library. related publication: OctoMap: An Efficient Probabilistic 3D Mapping Framework Based on Octrees 1)Octamap Volumetric Model 2)Probabilistic Sensor Fusion Model 3)Sensor Model for Laser Range Data
Octomap explanierend Read More »
Gaussian Mixture Regression is basically Multivariate normal distribution with Conditional distribution. The more about the theory could be found at [1], [2], [3], [4]. For this work, I have added the functionality of adding Gaussian Mixture Regression to this project on the GitHub by forking the main project, my forked project can be download at here Github The main changes
Gaussian Mixture Regression Read More »
I found a really good code at GitHub for fitting a Gaussian Mixture Model (GMM) with Expectation Maximization (EM) for ROS. There are so many parameters that you can change. Some of the most important ones are:
|
1 2 3 4 5 6 7 |
// minimum number of gaussians #define PARAM_NAME_GAUSSIAN_COUNT_MIN "gaussian_count_min" #define PARAM_DEFAULT_GAUSSIAN_COUNT_MIN 1 // search will terminate when the gaussian count reaches this #define PARAM_NAME_GAUSSIAN_COUNT_MAX "gaussian_count_max" #define PARAM_DEFAULT_GAUSSIAN_COUNT_MAX 10 |
To find the optimal number of components, it uses Bayesian information criterion (BIC). There are other methods to find
Expectation Maximization algorithm to obtain Gaussian mixture models for ROS Read More »
In this part, I trained an SVM over images of “face” or “not face” (36 × 36 pixels), using HOG features. I used VLFeat library for both HOG and the SVM. Example of face images: Example of nonface images: I divided the dataset into a training and a test set (80% and 20% respectively) and computed the HOG
Multi scale face detector using HOG features and support vector machine Read More »
This work is the second part of my master thesis (part I). In this part, I developed an algorithm for 2D pose estimation of the human body. To do this, I created a software with QT that could generate 2D contours representing human body. Then I send these contours for evaluation to CNS(Contrast Normalized Sobel) [1]
2D pose estimation of human body using CNS and PCA Read More »
This is the part I of the work that I did for my master thesis (part II). In this work first, I computed HOG (Histogram of oriented gradients) on my images and then sent the computed histogram to a linear SVM (support vector machine). The SVM was trained with human and non-human images. The output of the classifier was
Human detection on mobile camera using HOG and tracking them using Kalman filter Read More »
Real-time object recognition and 6DOF pose estimation with PCL pointcloud and ROS. This approach is based on normal coherence
Real time object recognition and 6DOF pose estimation with PCL pointcloud and ROS Read More »
In this tutorial, I’m gonna show you how to do object recognition and 6DOF pose estimation in real-time based on Linemod algorithm with ROS and PCL pointcloud. First, you need to install ork:
|
1 |
sudo apt-get install ros-indigo-object-recognition-ros-visualization ros-indigo-object-recognition-msgs ros-indigo-object-recognition-core |
Then add the model of your object for tracking to Couch DB:
|
1 |
rosrun object_recognition_core object_add.py -n coke -d "A universal can of coke" --commit |
|
1 |
rosrun object_recognition_core mesh_add.py <YOUR_OBJECT_ID_FROM_PREVIOUS_STEP> /home/behnam/ork_tutorials/data/coke.obj --commit |
You need to install Kinect driver, if you don’t
Gradient descent is a very popular method for finding the maximum/ minimum point of a given function. It’s very simple yet powerful but may trap in the local minima. Here I try to find the minimum of the following function: $$ z= -( 4 \times e^{- ( (x-4)^2 +(y-4)^2 ) }+ 2 \times e^{- ( (x-2)^2 +(y-2)^2
Gradient descent method for finding the minimum Read More »
Rapidly-exploring random trees (RRT) and their variant are a very power solution for solving motion planning problem in robotics, but they suffer from finding an optimise solution and the generated path is usually jerky with redundant movements. Sample based-optimisation-based planners benefit the robustness of RRT and the possibility of imposing a cost function. Here in this work, I
Sample based-optimisation-based planner with signed distance fields cost map Read More »
