Visual-Based Kinematics and Pose Estimation for Skid-Steering Robots

To build commercial robots, skid-steering mechanical design is of increased popularity due to its manufacturing simplicity and unique mechanism. However, these also cause significant challenges on software algorithm design, especially for the pose estimation (i.e., determining robot’s rotation position) since they change their orientation with an inevitable skid. tackle this problem, we propose a probabilistic sliding-window estimator dedicated using measurements from monocular camera, wheel encoders, optionally inertial measurement unit (IMU). Specifically, explicitly model kinematics robots by both track instantaneous centers (ICRs) correction factors, which are capable compensating complexity track-to-terrain interaction, imperfectness terrain conditions smoothness, etc. prevent performance reduction in robots’ long-term missions, time-and location-varying kinematic parameters estimated online along states tightly-coupled manner. More importantly, conduct in-depth observability analysis different sensors configurations paper, provides us theoretical tools making correct choice when building real robots. In our experiments, validate proposed method simulation tests real-world demonstrate that outperforms competing methods wide margins. Note Practitioners —This paper was motivated problem commonly commercial-used only low-cost sensors. Skid-steering skid, poses challenge encoders. We estimate kinematics, succeeds slippage, smoothness. It critical jointly missions. further theoretically analyze whether can be under sensor configurations, find out special degrade motions make unobservable.