Help with autonomous Robot using SLAM
Hello, I'm a Software Engineering student with zero background regarding Robots and for some stupid reason I took my final year project as a Autonomous Robot that navigate through the unknown environment and generate map. The problem i'm currently facing is the my project advisor who actually suggested us the project and have a robotics background is no help to us. He basically want us to do all the work which I'm willing to do but the problem is I'm having a very hard time figuring things regarding how I should start, what I should I look for, what should be the steps I should take to start the project once again I have zero knowledge about robots. We want to make a simulation and then make an actual prototype. I would really appreciate if any of you can guide me to the right steps, how I should start with this project what I should learn and what not. I just need help with that. I want to do all the work by myself just need some help with starting the whole thing.
Things I've done - Installed ROS Kinetic - Went through ROS tutorials - Made a service and a client - Went through Kinetic tutorials - Made packages, workspace - Used Gazebo - Made Environment using Gazebo - Made a Robot using URDF
Thank you so much.
Can you tell us what type of unknown environment you'll be mapping, buildings, forest, caves. It makes a big difference. Also do you know what sensors you can use? 2d lidar, 3D RGB-D camera, stereo camera, GPS. The sensors available will completely change everything else so need to be decided early.
The environment for mapping will be buildings and the sensor that we will be using are ultrasonic sensor. Thank you for the reply.
Take a look at SLAM algorithms and packages that can operate in these environments and try to use them with Gazebo and Rviz to make the map
I agree you will need to read up on a lot of SLAM algorithms for this. I've got to admit using just ultrasonic sensors is definitely going to make your life hard. LiDAR, camera, or RGB-D sensors would all be significantly easier.
I made a robot and then used it in the environment that I've made in Gazebo. I was actually having problem finding the ultrasonic sensor plugins for gazebo. I think I might have to drop the ultrasonic sensor. Can you please suggest any sensor that I can in simulation and is cheap so that I can use.
Yes, I actually looked into SLAM algorithm I think I'm getting close to the solution. Thank you again for the reply!
Lidar would be the best option and will give you the best results, but it's not that cheap
The Intel Realsense RGB-D sensors are probably the best balance between cost and value. They're just over $100 and you could use the RGB-D or RTABMAP ros SLAM packages out of the box. The kinect sensor plugin could be used in Gazebo to simulate the same type of data too.