Model-Based Control of High Speed Rough Terrain Robotic Vehicles

Principal Investigator

Steven Dubowsky

Group Members 

Dr. Karl Iagnemma, Research Scientist
Dr. Yoji Kuroda, Visiting Professor, Meiji University

Project Overview

High-speed mobile robots have many potential applications, including military reconnaissance and scientific exploration. Our project is focused on developing the control and planning algorithms for high speed autonomous rough-terrain ground vehicles. Currently, state of the art high-speed autonomous rough-terrain robotic vehicles travel at speeds of about 10 mph on relatively gentle terrain. Our goal is to develop a system for traversing rugged, desert-type terrain at substantially higher speeds, tolerating vehicle slip and ballistic motion.

Full-scale high-speed off-road military vehicle

Artist rendition of high speed robotic vehicle in rough terrain


  • Develop a high-fidelity, experimentally-validated model of a scale vehicle.
  • Develop physics-based control algorithms for high-speed rough terrain mobility using available sensor information
  • Demonstrate these control algorithms on a 1/4 scale vehicle in rough terrain.

Current Status

The focus of Year 1 for the project was to begin initial development of a low-level control scheme, and to develop an experimentally validated model of a representative high-speed rough-terrain vehicle system.  A small tele-operated vehicle was built for this purpose.  The vehicle dynamics were identified, modeled, and subsequently simulated using a commercially available dynamics solver package.  The simulation results compared favorably to the experiments.  Below are images from our rough-terrain vehicle simulation.

front iso wireframe white.jpg (96701 bytes)

Wireframe view of simulation model

Rough terrain vehicle simulation video (830 KB)

We are also investigating various low-level control schemes in simulation, and in the future, experimentally.  These schemes will be used for velocity and heading control in rough terrain at high speeds.  Low-level control will be combined with other control and planning layers to enable robust, fully autonomous behavior. 

Experimental Systems

We have built two experimental systems for this project a small 1/10 scale vehicle for model validation experiments, and a larger, 1/4 scale vehicle for autonomous control experiments.


MOD 1 vehicle

iso front shaded.jpg (35364 bytes)

Simulation of MOD 1

Above is a picture of the 1/10 scale system built modeled for validation purposes.  The vehicle is tele-operated, and fitted with a three axis accelerometer and data-acquisition equipment for recording the dynamic response.  The two-wheel drive vehicle is commanded torque to the drive wheels, and front wheel steering angle.  The wheels are driven by an electric motor through a rear differential.  The vehicle is about 14 in long, and weighs about 3 kg.

Video of MOD 1 in action in the woods (2.8 MB)


artemis 2.jpg (64039 bytes)

Artemis "undressed" on the astroturf

Artemis internal schematic

We are building a 1/4 scale high speed rough terrain testbed called Artemis for testing and demonstrating control and planning algorithms in real time.  Artemis is about 48 inches long, current wheel diameter of about 7 inches, and weighs about 15 kg.  The vehicle chassis, suspension, and drivetrain were custom made to our specifications by Weller Racing.   With no payload, the vehicle can travel up to 50 mph, and survive drop offs of over 6 feet.  Components and systems:

  • 4.5 hp 2 stroke gasoline powered Zenoah G2D70 engine with custom clutch
  • Futaba digital servo actuation of steering linkage and throttle body
  • EEPD Profive 700 Mhz Pentium III PC -104 onboard computer running Real Time Linux
  • Data acquisition system
  • Custom built servo control and system security electronics
  • Crossbow AHRS-400 inertial measurement system
  • Differential GPS
  • Radio modem
  • Videre Design Stereo Pair
  • Battery cluster


    Defense Advanced Research Projects Agency (DARPA)

    US Army Tank-automotive and Armaments Command (TACOM)



Copyright 1995-2009 by MIT.

All rights reserved.