Real-Time Physics Simulation Forum 2021-01-18T11:06:05+00:00 https://pybullet.org/Bullet/phpBB3/app.php/feed/topic/13138 2021-01-18T11:06:05+00:00 2021-01-18T11:06:05+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=43255#p43255 <![CDATA[Physics error for wheel with TORQUE_CONTROL]]> I want to use TORQUE_CONTROL for driving the wheels, because the torque is what I set when
setting the PWM on a real motor controller. I was able to do a control loop that stabilizes my robot.

But the final state does not capture the correct physics. The robot is standing at a fixed place
leaning to one side and applies torque to its wheels to counteract the gravity force that wants to tip it over.
The wheels are spinning so the friction between the wheels and the plane just balances the torque.
This should never be a stationary state! The robot must accelerate to the side - it doesn't !
It stands still leaning 5 degrees to one side!
pybullet_error.png

Interestingly, it works with VELOCITY_CONTROL, only the toque jumps all over the place.

The URDF snippet descibes one wheel:

Code: 

  <link name="right_wheel">    <visual>      <geometry>        <cylinder length="1.0" radius="3.5"/>      </geometry>      <origin rpy="0 1.57 0" xyz="0 0 0"/>      <material name="red"/>      <origin rpy="0 0 0" xyz="0 0 0"/>    </visual>    <collision>      <geometry>        <cylinder length="1.0" radius="3.5"/>      </geometry>      <origin rpy="0 1.57 0" xyz="0 0 0"/>    </collision>    <contact>      <lateral_friction value="0.5"/>      <rolling_friction value="0.01"/>      <stiffness value="100000"/>      <damping value="3000"/>    </contact>  </link>

Statistics: Posted by lehnertu — Mon Jan 18, 2021 11:06 am

]]>