The MULE: Multipurpose Logistics Vehicle

Anatomy of a U.S. Army Warrior Bot

by Francis X. Govers, Chief Engineer for Elbit Systems of America

Regardless of your feelings for the quality of the writing or storyline, you have to admit that the TV program “Knight Rider” has helped to plant the concept of a self-driving car into the public’s consciousness. Now that KITT, the self-driving talking car, has returned to the screen in 2008, and with the successes of robot cars in the DARPA Urban Challenge (see Ken Berry's report in our Spring 2008 issue), autonomous cars seem far more plausible, and indeed, inevitable. As a current baseline of where we are today versus the ideal of KITT’s intelligence, we can compare the smart-aleck Mustang GT500 to a current production Unmanned Ground Vehicle (UGV), the U.S. Army’s MULE, or Multipurpose Logistics Vehicle, which is nearing testing and production. The MULE is being built under contract by Lockheed Martin in Grand Prairie, Texas, a company more famous for building fighter jets.

Any robot sees the world through its sensors. UGVs need to be able to sense a wide variety of objects—cars, the terrain, trees, people, buildings, weaponry—and have a variety of sensors to do this job. They also need to understand where they are and where they are going.

First of all, pretty much every UGV uses some form of the Global Position System (GPS) and has on board some type of mapping system that understands roads and terrain to position the vehicle and to plan ahead for the route it is to take. The operators of the UGV assign it “waypoints,” or spots on the map, that describe the route it is to take. While a robot aircraft or unmanned aerial vehicle (UAV) can blindly follow GPS steering instructions, obviously, a ground vehicle cannot.

3D Ladar

The main obstacle-avoidance system on board is the 3D LADAR (“Laser Radar”) system. How does it work? Imagine a madly spinning mirrored disco ball illuminated by a bright laser and throwing off beams in all directions. Each beam as it sweeps around hits objects or the ground, and is reflected back to the sensor. By measuring the very short time it takes for the light to flash out and back, the LADAR builds up a 3D view of the area around the vehicle, and it is accurate to fractions of an inch. It is a 3D laser scanner making dozens of sweeps a second.

The onboard computers go through this data and classify the areas around the vehicle as ground (mostly flat), road (flat areas with curbs), trees (tall round things with a halo of leaves), other vehicles (boxy things that move) and people (people-size things that move). Everything else is an obstacle to be avoided. The computer could assign a “mobility cost” to each area, for example, from 0 for a nice, flat road to 10 for bumpy, rough regions and 1,000 for steep, impassable areas or areas with too many obstacles. A path-planning system finds the route through the area that has the least “cost,” and this result is the most efficient path through the terrain.

Occupancy Grid

One of the real technological challenges in navigating a UGV is to combine the obstacle data from the LADAR with the GPS data to divide the world into moving objects and non-moving ones. This is not a simple challenge, as the vehicle is moving all the time, bouncing over ruts and potholes and turning to avoid obstacles. A very precise Inertial Navigation System (INS) is required; this comprises a series of accelerometers (that measure movement) and gyroscopes (that measure turning and tilting). A lot of very clever computer programming reorients the vehicle, the map and all of the obstacles many times a second to keep them all in sync. Robotics engineers call this resulting map an “occupancy grid,” since it divides the world into areas that are occupied by obstacles and areas that are clear.

Multi-Sensor Fusion

As useful as the LADAR is, it does not tell the whole story. One property that can be used to distinguish people and other vehicles from the general group of obstacles is the presence of heat. Humans radiate heat at 98.6 degrees, and cars and trucks are considerably hotter. After all, we need our UGV to distinguish between a log (that it can safely drive over) and a person lying down (which it cannot). The MULE UGV has multiple cameras mounted all around it, providing vision in several modes that include normal color and infrared. As an Army vehicle, the MULE must operate on almost any terrain in any weather.

Other sensors include a combat ID system that tells friend from foe and an acoustic locater array that can tell the remote operator of the vehicle where small-arms fire is coming from.

Telepresence Firepower

There are three variants of the MULE under development. The Transport MULE is a truck-like vehicle that is designed to carry supplies and equipment. The Counter-mine MULE locates buried land mines with ground-penetrating radar and can neutralize those mines. The final MULE is the Armed Robotic Vehicle —Assault (Light), or that carries anti-tank missiles and a heavy machine gun. While the MULE is fully capable of driving itself, it has no ability—none whatsoever—to fire its own weapons. This can only be done by the vehicle’s operator.

Power, Brains & Versatility

One of the most surprising aspects of the MULE UGV is its hybrid electric/diesel powerplant. Each of the six wheels contains an electric hub motor that actually turns the wheel. The diesel engine is connected only to an electric generator. This configuration allows the suspension to rotate the suspension arms through a large range, permitting an amazing maneuvering capability, including the ability to climb over some barriers and cross gaps. The MULE can also lift any two wheels off the ground and still drive or put its body on the ground with all six tires in the air to reduce its profile or to allow maintenance. As part of the Army’s Future Combat System, the MULE is designed to be maintained with only 10 tools, which is sure to be a big relief to future Army robot mechanics.

These electronics and mechanics are managed by a small cluster of Linux-based real-time computers. There are three sets of computers: the Autonomous Navigation System (ANS), the Vehicle Management System (VMS) and the Battle Command system that performs mission planning and talks to the rest of the Army’s systems.

MULEs are designed to fit seamlessly into existing Army doctrine, which means that they must be able to move in formations or in convoys. Any MULE vehicle can be a formation leader, and all of the unmanned vehicles can follow in formation or in a convoy train. Given all this technology, the current UGV capability is coming up hard to match the fictional abilities of KITT, making our Army troops the “knight riders” of the future.

Francis X. Gover is a Chief Engineer for Elbit Systems of America. Elbit makes land robots, UAV's, and USV's (Unmanned Surface Vehicles - i.e. autonomous boats), as well as all sorts of displays and sensors. -the Editors

Links

Future Combat Systems www.fcs.army.mil
Wikipedia on Future Combat Systems www.wikipedia.org
Boeing www.boeing.com/defense-space
Global Security www.globalsecurity.org

Return to the home page