The minimec is a custom mecanum wheel base made for development. This repository contains a simple C++ library and ROS2 packages that make the software running on the robot.
Although this was purposely done for the Minimec, I tried to keep the software decoupled from the hardware. A user could ideally swap out the minimec_driver package for one that implements communication with their own hardware and still have mostly everything working.
minimec_bringup- necessary launchfiles and configuration to get the minimec up and running quickly.minimec_control- all things controls. Kinematics, odometry, path generation, trajectory tracking.minimec_description- robot definition, URDF, meshes.minimec_driver- interface with motion hardware (odrives). Although the odrive-provided nodes are used to communicate with the odrives over CAN, this provides an extra layer of abstraction.minimec_msgs- custom messages and services for minimec operation.minimec_lights- control of Neopixel LEDs mounted on the minimec.
Each package contains a README that goes into more detail.
The MecanumDrive class is the core of this library.
It provides the following functions:
- FKin - Forward kinematics, maps wheel positions to transforms.
- IKin - Inverse kinematics, maps input twist to wheel speeds.
- setWheelOffsets - set initial encoder offsets
Four nodes form the odrive_can package communicate the onboard computer with the Odrives. The minimec_driver node serves as a hardware abstraction node, publishing and subscribing to commands and encoder feedback. The odometry node calculates odometry based on the Odrives' feedback. The trajectory_tracker tracks a path using Feedforward + PI control, outputting velocity commands. The kinematics node translates velocity commands into speeds needed by the Odrives. The minimec_lights controls the LEDs based on /cmd_vel. The path_generator offers services that let the user generate paths, which will be tracked by the trajectory_tracker.
The same workspace should be on the robot and your computer.
Compiling for the robot:
colcon build --cmake-args -DBUILD_TESTING=OFF -DBUILD_ROBOT=ON, the -DBUILD_TESTING flag is optional, but it prevents from compiling tests. Note: cross compilation is recommended, as compilation on the Pi can take quite a bit.
Compiling for the control computer:
colcon build --cmake-args -DBUILD_TESTING=OFF -DBUILD_ROBOT=OFF
Make sure the robot and your computer are on the same network, and share the same ROS_DOMAIN_ID.
On the robot:
ros2 launch minimec_bringup launch_minimec.launch.py cmd_src:=teleop - this will launch the necessary nodes on the robot, and the robot just listens to the topic /cmd_vel and moves accordingly (no feedback). You can use the teleop_twist_keyboard node to drive the robot around.
On the control computer:
ros2 launch minimec_bringup launch_command.launch.py cmd_src:=teleop - this will launch the teleop node on a separate terminal.
For launching the robot to track paths.
On the robot:
ros2 launch minimec_bringup launch_minimec.launch.py cmd_src:=path - this will additionally launch the trajectory_tracking node. Whenever a path is published, the tracking node will start following.
On the computer:
ros2 launch minimec_bringup launch_command.launch.py cmd_src:=path - this will launch rviz and the path generation node.
Note: be mindful of the current robot location when you run the trajectory tracking. If the robot is too far from the first point in the path, it will try to make the correction pretty quickly.
Refer to the readme in the minimec_control package to see how to use this mode.
A /setup folder is included in this repository to help set up the robot easily and quickly. This is already done, but can serve as reference if there is a need to reflash the robot's storage.
- Follow the steps in
/setupto cofigure the robot.