This is the ROS package of the control architecture for Nermo.
Before attempting to runs the mouse_controller, ensure that ROS is installed, a catkin workspace is set up. For some initial guidance, follow the tutorials offered by the Open Robotics community here.
The ROS package is currently configured to take inputs from an upstream connected controller (e.g. a PS4 controller) using the ROS Joy package.
Therefore, before running the mouse_controller, ensure that a controller is connected and the joy node active. Otherwise, the mouse will load, but no movements happen.
To run the mouse_controller, simulation etc. launch using
roslaunch mouse_controller mouse_controller_launch.launch
The mouse_controller also works for basic straight-line tests without a controller. To achieve this, currently some minor modifications need to be made in the high_level_controller.py and motion_module.py.
Changes in high_level_controller.py
Uncomment the line in the main method loop #vel_in = 0.4. This parameter can be freely adjusted to change the velocity of the mouse. Essentially, the controller now sends a continous, constant control signal to the motion module.
Changes in motion_module.py
In standard configuration, the mouse is in a non-moving neutral configuration. To ensure the mouse walks without a control input, change the parameter self.balance_mode from True to False.
On a high level, the control architecture of Nermo combines quadruped legged motion control and spine motion control. These two motions are not controlled in isolation and therefore the control architecture ensures synchronization between the leg and spine control.
The diagram below provides an overview of the ROS implementation of the control architecture. The diagram shows the different control nodes, message topics, as well as the connection with the physical mouse.