Robotics Stack Exchange is a question and answer site for professional robotic engineers, hobbyists, researchers and students. It only takes a minute to sign up.
Sign up to join this community
I am migrating from a differential drive design to a skid steering design for my robot, and I want to know how easy would it be to use the NavStack with skid steering. Would there be any problems in terms of localization and things like that?
If I let two wheels on the same side of my robot (two on left side and two on the right side) maintain same velocity and acceleration, would the unicycle model of a differential drive robot still apply for skid steering?
Yes and no. I think it depends on your needs, and what other sensors you have.
Remember that even diff drive robot wheels slip a little. And you can never measure the diameter of the wheel accurate enough. This is why dead-reckoning doesn't work well over long distances. And why you need other sensors to correct for these errors.
I haven't used the entire ROS Navigation Stack, but I have used parts of it with a skid steer robot. (In my case I used a Hokuyo, robot_pose_EKF, laser_scan_matcher, and costmap_2d nodes among other custom nodes.) My setup was enough for the robot to maintain a reasonable 2D location if it didn't drive too far. But after a long time of driving around, the location and orientation would be way off. I don't think doing any sort of SLAM mapping would have worked for me. I found that the ICP algorithm in the laser_scan_matcher was critical to correcting the odometry errors when the robot would turn.
I should mention that I did not have an IMU to feed into the EKF, but i think that would have improved results greatly. The diameter of my wheels was also way off. (I would drive a meter, but the robot would think it was only 0.8 or so.) So if you have better odometry, a laser, and an IMU, then you might be ok. But i don't know what your application is...
