agbj's profile - activity

I saw that we can signal move_base that the path is blocked in order to recalculate the global plan as quoted in the next sentence.

~planner_frequency (double, default: 0.0)

The rate in Hz at which to run the global planning loop. If the frequency is set to 0.0, the global planner will only run when a new goal is received or the local planner reports that its path is blocked.

My question is: what is the mechanism in the local planner to inform move base that the path is blocked?

Another question that comes to mind is the centralized architecture problem : If everything is controlled by the ROS master what happens if this entity fails? Does the ROS framework have control mechanisms to recover or stop everything if there is nothing that can be done to recover?

Inform about blocked plan I saw that we can signal move_base that the path is blocked in order to recalculate the global

Thank you. And if the master fails during the connections establishment?

Inform about blocked plan I saw that we can signal move_base that the path is blocked in order to recalculate the global

Inform blocked plan and rosmaster fail I saw that we can signal move_base that the path is blocked in order to recalcula

Thank you. One question answered.

Thank you. One question solved.

Inform blocked plan and rosmaster fail I saw that we can signal move_base that the path is blocked in order to recalcula

Inform blocked plan and rosmaster fail I saw that we can signal move_base that the path is blocked in order to recalcula

The problem is in your CMakeLists. Either you don't search for the required package in find_package macro or you don't i

My approach to simulate a car like robot was to develop a SDF model because it has a revolute2 joint that allow the wheels to rotate around two different axis simultaneous. In my case URDF is made just for RVIZ visualization and to define the transformations between robot's frames. Hope it helps. Any doubt feel free to ask

To compare between odometry and the real position you have to develop a node that compares the odometric information with the information given by the plugin that Stefan Kohlbrecher mencioned

gmapping is a SLAM node. If configured properly it will give you the transform that you pretend and build the map that you can store later. If you have already a map you can run only a localization node like amcl that also gives you the transform between map and odom.

This odometry node publish the transform between the odom frame and the chassis_link as well. So, from what you said i think that you need a localization node as well as a map. Think about it before pass the code to the real robot. Any doubt just ask

Let me explain the process: Normally the map frame appear when you have a localization node like amcl that publishes a transform from the map frame to the odom frame. The odom frame is typically created by the node that publish the odometry. Next comment please

The frame between map and odom is being published by the ukf node, but it is incorrect because in rviz the robot just stands still and i think it is because I am publish both the pose and odometry in the odom frame. When I use only my localization node the robot is able to localize itself reasonably

What you want is the real pose of the robot that gazebo can give you. That being said, the position that the robot think it is is given by the transform between the map and the base_link frame.

Normally in a scenario with localization you have at least two frames that are the map and base_link frame and if you are using odometry i have the odom frame as well. The map frame is fixed in the world and the base_link is fixed to your robot. Read the next comment please

Thank you for the answer but to publish both the pose and odometry in the same frame to get the transform from map to odom. I think i have to publish the odometric information in the map frame. Correct me if I am wrong. But the node that publishes the odometry does not know how to transform the pose

You can create a model plugin for jackal. Through this plugin you can easily extract the real pose of the model pose and publish this information to a ROS topic.

See this link: gazebo model plugin

A SDF model can only be used inside gazebo. Rviz and other ROS related software uses only URDF. However you can add to your URDF gazebo tags that allow your URDF model to be deployed inside gazebo because this simulator converts, under the hood, your URDF model to an SDF model.

First of all try to use std::vector instead of raw pointers. It is best practice to do so. Your code is in c-style, not cpp style. Your problem can be solved doing what i have mentioned.

int** pointCloud; //double pointer for 2D Array with dimension sizeX and sizeY
std::vector<uint8_t> rasterImage; //array-variable
int sizeX; //amount of pixel along X-axis
int sizeY; //amount of pixel along Y-axis
int index; //represents value of reflected power

pointCloud = new int* [sizeX];
for (int x = 0; x < sizeX; x++)
{
    pointCloud[x] = new int [sizeY];
    for (int y = 0; y < sizeY; y++)
    {
        int index = (sizeX * y) + x;
        pointCloud[x][y] = rasterImage[index];
    }
}

for (int y = 0; y < sizeY; y++) { for (int x = 0; x < sizeX; x++) { printf("%d ", pointCloud[x][y]); } printf("\n"); }

sensor_msgs::Image depth_image;

depth_image.height = sizeY;
depth_image.width = sizeX;
//depth_image.encoding = rgb8;
depth_image.is_bigendian = false;
depth_image.step = sizeX * sizeY;
depth_image.data = rasterImage;

Typically the .launch files stay inside a catkin package in the folder launch. Not sure if this is the answer that you pretend. If not please be more explicit

I never mixed python 3 with ROS but do you tried to install python3 through the repositories and not through anaconda?