Wheeled Mobile Robot Dynamic Modeling
This topic is the core of my Ph.D. thesis research (see my proposal). Wheeled mobile robots (WMRs) require motion models for position estimation, control, and planning. Most fielded WMRs today use basic 2D kinematic models, but these are inaccurate. I am developing high-fidelity WMR dynamic models that account for 3D joint articulations, inertial effects, wheel slip, and actuator characteristics. My formulation is modular to accommodate any joint configuration and any nonlinear model of wheel-terrain interaction. My formulation is also computationally efficient such that models can be forward simulated much faster than real time.
I am also investigating model reduction techniques. I have shown that 3D kinematic models can be enhanced to account for the effects of wheel slip on body-level motion. In many cases, these provide comparable accuracy to full dynamic models at a fraction of the computational cost. I have also shown that our IPEM method can be used to calibrate 3D kinematic and dynamic models. The goal of my modeling research is to push the limits of WMR mobility. Only with high-fidelity models can WMRs navigate over very rough terrain and at very high speeds. I plan to release an open source modeling/calibration library that will provide unique functionality and faster computation compared to currently available resources for WMR simulation (e.g. Open Dynamics Engine, CarSim).
Below are just a few example WMR models. Using my software library, constructing and simulating these models requires just a few lines of code.
Alonzo Kelly and Neal Seegmiller, A Vector Algebra Formulation of Mobile Robot Velocity Kinematics, Proc. Field and Service Robotics, July, 2012. |pdf|
Neal Seegmiller and Alonzo Kelly, Modular Dynamic Simulation of Wheeled Mobile Robots, Proc. Field and Service Robotics, December, 2013. |pdf|
This research was made with U.S. Government support by the Army Research Laboratory (W911NF-10-2-0016) and by the National Science Foundation Graduate Research Fellowship (0946825)
Copyright © 2014 Neal Seegmiller