Steering for sandwich and CoRa not as expected

[matthewheubach]

Steering response for sandwich boats is poor.

Looking at the rostopic list, it appears there is an issue with the way the left and right thrust_angle topics are initialized (the namespace is wamv instead of sandwich_0 or sandwich_1)

/sandwich_0/joint_states
/sandwich_0/thrusters/left_thrust_cmd
/sandwich_0/thrusters/right_thrust_cmd
/sandwich_1/joint_states
/sandwich_1/thrusters/left_thrust_cmd
/sandwich_1/thrusters/right_thrust_cmd
/vrx/debug/wind/direction
/vrx/debug/wind/speed
/wamv/shooters/ball_shooter/fire
/wamv/thrusters/left_thrust_angle
/wamv/thrusters/right_thrust_angle

[bsb808]

@matthewheubach Can you explain in more detail. This is good practice.

Need to know the following:

• What is the setup you used, i.e,. what instructions did you follow or what other commands did you do to run the test?
• What output did you expect?
• What output did you observe - and how did that differ from what you expected?
• What is the next step? E.g., are you looking for feedback, would you like someone to try something (tag @crvogt or @bsb808 ), do we just leave this for later?

[matthewheubach]

@bsb808 @crvogt

To test the steering response I used the sandwich_keydrive.launch file (added to the sandwich_gazebo/launch folder). I developed this basic keydrive functionality based on the cora_keydrive.launch file developed in MRC. I tested the keydrive capability on the cora vessel to ensure functionality and then changed the thruster namespace to sandwich_1 for sandwich boat control.

After opening the sandwich_keydrive.launch application, I used rostopic echo to display the gps speed and imu angular turn rates.

Linear velocity is around 3 m/s which would be reasonable for the life-size boat but significantly slow for the 10x model. The maximum turn rate I could get was 0.029 rad/sec. The model boats have a real-life capability of at least 45 deg/sec (~0.8 rad/sec).

If we use the model at the 10x scale, the thrusters might need to be re-configured to produce similar behavior to the small-scale model in terms of turn rate and speed (approx. 1.5-2 boat lengths/sec)

[bsb808]

@matthewheubach There are a couple of interrelated settings that control the boat motion:

• The inertial parameters are set for the links that makeup the boat/robot. These parameters are used by the Gazebo physics engine. See https://github.com/Field-Robotics-Lab/sandwichboat/blob/main/sandwich_description/urdf/sandwich.urdf.xacro
• We add the hydrodynamic parameters are defined in a separate plugin.
  • The values of the parameters are defined in this xacro file: https://github.com/Field-Robotics-Lab/sandwichboat/blob/main/sandwich_description/urdf/sandwich_dynamics_plugin.xacro
  • I shared an overleak project with you that documents the model
• The propulsion behavior has a couple of elements:
  • The location of the thrusters is defined here: https://github.com/Field-Robotics-Lab/sandwichboat/blob/main/sandwich_description/urdf/thrusters.xacro See particularly the pose values of the right_chasis_engine joint and of the left_chasis_engine_joint. The port/stbd separation appears to be +/- 0.7 m (the y component of the pose) which will dictate how much torque is generated by the differential thrust.
  • The amount of force generated by the thrusters is defined here: https://github.com/Field-Robotics-Lab/sandwichboat/blob/main/sandwich_description/urdf/sandwich_thruster_config.xacro See the maxForceFwd and maxForceRev values. The mappingType 0 is linear interpolation between zero->max values. This plugin comes from VRX and there is some documentation in the plugin header file: https://github.com/osrf/vrx/blob/3408b8bbe3d794dbcdee14241dea57d9f8307f6d/usv_gazebo_plugins/include/usv_gazebo_plugins/usv_gazebo_thrust_plugin.hh#L113