Modeling and Simulation of an Unmanned Ground Vehicle Power System

These models can be used to plan missions or roughly estimate power system operation for an unmanned ground vehicle.

Robotic vehicles such as unmanned ground vehicles (UGVs) have multiple sources of power, including batteries, fuel cells, combustion engines, ultracapacitors, and solar cells to allow for extended periods of operation. Fuel-based power sources have a higher specific energy than batteries, which is why most current automobiles are gasoline-powered. Batteries have many other advantages in terms of low noise profile, easy replacement, and direct energy conversion. Solar charging allows for harvesting of natural resources to increase total energy reserves. Mission duration may be maximized using a combination of power systems.

The fuel cell connected to a TALON robot battery pack.

To effectively integrate multiple power system components, a modeling framework was developed to simulate and plan operation of UGV power systems. First, each component is individually modeled using either empirical or theoretical techniques. These models consider power component states such as time of operation, state of charge, and temperature. For a given mission, the power demand is estimated and the power system models are combined to compute total energy use. As a part of the model, energy losses due to the operation of power system components are accounted for. Losses include resistive heating in batteries, and startup or shutdown power demands. In addition to full, nonlinear models for the UGV power system, a simplification process was demonstrated that can be used to reduce the models to linear dynamics. These simplified models can be used to plan missions or roughly estimate power system operation for a desired mission.

The fuel cell used in this work can only be turned on or off, with no variation in the power produced when on, and requires several minutes and nontrivial power input to transition between on and off. These limitations on the fuel cell lead naturally to the proposed hybrid systems framework.

The power system model consists of the 200W fuel cell and a Li-ion battery. The fuel cell is fueled by commercial propane canisters and consists of a 200W solid oxide fuel cell, a fuel reformer, and a DC/DC converter. The propane gas is first desulfurized and then reformed via partial oxidation into a hydrogen- rich fuel stream to feed the fuel cell. The fuel cell was designed to be integrated with existing batteries on small UGVs such as the TALON robot. This combined power system significantly increases the possible mission duration, especially under low-power loads such as persistent stare missions.

One of the challenges of integrating this power source is to develop an optimal duty cycle for using the fuel cell to recharge the batteries. The fuel cell was connected to a TALON battery pack with a moderate state-of-charge (SOC) and issued a startup command. The current draw from the batteries was logged every 10 seconds until the fuel cell completed the startup procedure and began charging the batteries. Likewise, the fuel cell was then issued a shutdown command, and the current draw was logged every 10 seconds until the fuel cell shut down. Power and energy values were calculated using the average voltage of the battery back (35 Volts) throughout the tests. The fuel cell consumed approximately 6.5 Watt-hours over 16 minutes to start up, and approximately 5.3 Watt-hours over 18 minutes to shut down.

To determine the power requirements for the mission, a simulation model was used for the UGV operating in a known environment. This model includes motor models and track-terrain interaction models. From this model, one can simulate the desired mission and obtain the power demand over time for a given mission. The driving loads can be decomposed into resistance due to terrain and changes in kinetic energy.

To provide appropriate torque inputs to the terrain model, a motor model was obtained experimentally by testing iRobot Packbot motors. The model takes in the current shaft speed and the power being delivered to the motors, and calculates the torque output. Together with a simple rigid body model of the UGV, one can simulate the UGV completing a mission and record the power used. In addition to variable power demands due to locomotion, electronic components onboard require power for operation. It is assumed that these loads are known and constant over the entire mission.

Studying a simple fuel cell/battery hybrid, this framework can be used to evaluate the performance of different control laws for desired criteria. In particular, energy losses of the entire power system and thermal response of the battery for an extended UGV mission were investigated. Future work includes using this combined model to plan and optimize energy efficiency for a mission. These models can be validated by running physical experiments with a UGV carrying a fuel cell and batteries running the controllers described.

This work was done by Jack Hartner of Army RDECOM-TARDEC; and John Broderick, Dawn Tilbury, and Ella Atlkins of the University of Michigan. ARL-0176