This is the repository which contains the workspace of the first project of the course Robotics at Politecnico di Milano.
University: Politecnico di Milano, Laurea Magistrale (MSc) - Computer Science and Engineering
We had to develop this project with ROS, and we had some bags recorded. The robot with which were recorded, was an omnidirectional robot with mecanum wheels (4 wheels with rollers at 45°). We had encoders on each wheel, which were giving to us RPMs (N.B.: they were not classic RPMs, in this case the acronym meaning was Radiants Per Minute), and Ticks of the encoder. The RPMs were really noisy, while ticks were more accurate. We were given the wheels radious, the wheel position along x (half length), and the wheel position along y (half width). Moreover, we had a ground truth pose of the robot acquired with OptiTrack.
The goal was to:
- Compute odometry using appropriate kinematics
- compute robot linear and angular velocities (v, omega) from wheel encoders;
- compute odometry using both Euler and Runge-Kutta integration
- ROS parameters for initial pose
- Calibrate (fine - tune) robot parameters to match ground truth
- Compute wheels control speeds from v, omega;
- Add a service to reset the odometry to a specified pose (x,y,theta)
- Use dynamic reconfigure to select between integration method (using an enum with two values: Euler, RK)
In particular, in point 1 we had to publish v, omega as topic cmd_vel of type geometry_msgs/TwistStamped, and once computed odometry we had to publish it as nav_msgs/Odometry on topic odom, and broadcast TF odom->base_link.
In point 2 we hade to compute wheels speeds (RPM) from v, omega, and publish them as custom message on topic wheels_rpm. The custom message needed to have as prototype:
Header header
float64 rpm_fl
float64 rpm_fr
float64 rpm_rr
float64 rpm_rl
where f/r stands for front/rear, and l/r for left/right. Then we had to check that the results matched the encoders value, apart from some noise.This is useful to check that the results match the recorded encoder values (using for example rqt_plot or plotjugler)
Initial data:
- Wheel radious
- Wheel position along x
- Wheel position along y
- Gear ratio
Given bags:
- bag 1 with only linear motions (no rotations)
- bag 2 with only forard motions and rotations (no y translation)
- bag 2 with a "freestyle" recording
- Luca Alessanrini
- Massimo Buono (added as collaborator)
- Mattia Portanti (added as collaborator)
We decided to face this challenge for the most part with in-presence meetings, where we discussed and implemented together different choices
We developed this on Ubuntu 18, with ROS Melodic. Once having downloaded the workspace, it is necessary to link it into the bashrc file and build it in order to use the project.
We had to submit not the whole workspace, but just only our package.