GetForceTorque (Force Torque Sensor) in robot wrist measures incorrect forces

osrf-migration commented 10 years ago

Original report (archived issue) by Guzmán Borque (Bitbucket: gborque).

This is my first issue report, please forgive me if I have made something wrong.

I have been developing a Force Torque sensor plugin for ROS Hydro and Gazebo 1.9, and I have encountered several issues about the readings of the sensor.

The plugin reads the wrench in a specific joint and publishes it to a ROS topic. The readings are obtained with physics::JointWrench GetForceTorque(0) method. The physics engine used is the default, ODE.

These issues are:

When an update rate of 1000 Hz is specified, when the robot is not moving, the FT sensor measures vary from +1N and -1N.
When specifying lower update rates, after starting the simulation, the sensor readings show a force of 1N in a direction parallel to the floor. If i exert an external force in that direction, the sensor measures the exact force that i have applied, there is no 1N extra. After stopping exerting that force, the sensors indicates no force for several milliseconds, but finally that force of 1N appears again.
The only way that I have found in order to prevent that force from appearing is moving the robot.

I am not sure what the source of this strange behavior is. If you want any more details about these issues, feel free to ask.

osrf-migration commented 10 years ago

Original comment by Guzmán Borque (Bitbucket: gborque).

Edited issue description

osrf-migration commented 8 years ago

Original comment by Nate Koenig (Bitbucket: Nathan Koenig).

set version to "all"

osrf-migration commented 6 years ago

Original comment by Luke Ramos (Bitbucket: roboticwalrus).

Is there a solution to this issue? I am aware that I am reading this years later. But I'm also trying to figure out an issue in my own simulation regarding a force torque sensor (that i did not develop. Please see link to package repo below). My issue involved a noisy signal when the sensor is in contact with an object. I have already raised this issue on github on the sensor repo. The noise is a fluctuation between a magnitude force reading and zero. I'm assuming this isn't the same error, but I just wanted to make sure that gazebo is indeed simulating collisions properly.

https://github.com/CentroEPiaggio/force-torque-sensor

osrf-migration commented 4 years ago

Original comment by Jack Liu (Bitbucket: jacknlliu).

Any update?

gazebosim / gazebo-classic

GetForceTorque (Force Torque Sensor) in robot wrist measures incorrect forces #1229