3^rd TACOM Robotics Quarterly Workshop

Advances In Autonomous Mobility in Complex Environments

sponsored by Joint Center for Robotics (JCR)

When: Monday, Apr 7,  2008

Where: U.S. Army TARDEC, Warren, MI

Agenda

Activity	Presenter
Advances and Opportunities in Autonomous Mobility in Complex Environments	Charles Reinholtz Embry-Riddle Aeronautical Univ.	view abstract	download presentation
TerraMax: Team Oshkosh Urban Challenge Robot	John Beck  Oshkosh Truck Corporation	view abstract	download presentation
GPS-free Tracking of Vehicles and Dismounted Personnel in Complex Environments	Johann Borenstein University of Michigan	view abstract	download presentation

Advances and Opportunities in Autonomous Mobility in Complex Environments

Charles Reinholtz, Embry-Riddle Aeronautical University

While tremendous advances have been made in the theory and application of autonomous vehicles in recent years, many challenges remain. Among the most important and significant challenges is safe navigation in complex environments, such as a busy city street or in a crowded shopping center. Successful technology demonstrations, including the DARPA Grand and Urban Challenges have shown promise for autonomous systems, but these experiments were conducted in relatively well-controlled and structured environments.

Recent advances and an improved understanding in the areas of sensor technology, navigation strategies, computational requirements, modular subsystems, and interoperability will be discussed, along with opportunities and needs.

back to top

GPS-free Tracking of Vehicles and Dismounted Personnel in Complex Environments

Johann Borenstein, University of Michigan

This talk presents an innovate and unconventional method related to the tracking of ground vehicles and dismounted personnel. The proposed method is useful in GPS-denied complex environments, such as indoors, underground, underneath dense foliage, or in the so-called urban canyon. When GPS is unavailable, the tracking of ground vehicles requires the computation of heading, usually from data provided by Inertial Measurement Units (IMUs), and the computation of distance traveled (i.e., odometry), usually obtained from wheel encoders. For the more difficult problem of tracking dismounted personnel, our group has developed the Personal Dead-reckoning (PDR) system, which uses a foot-mounted IMU and sophisticated software to produce rather accurate odometry data (but less accurate heading data) for walking persons. The PDR system will be discussed briefly, but it is not the focus of this talk.

Given that reasonably accurate odometry data from wheel encoders for vehicles and from the PDR system for walking persons is available, our focus has now shifted to the reduction of heading errors. Heading errors are currently the main bottleneck in GPS-denied vehicle and personnel tracking applications, especially since even small heading errors result in large position errors when travel distances are large.

The foremost source of errors in IMU-derived heading data is the drift of the gyros in the IMU. Other error sources in certain gyros, such as sensitivity to acceleration and temperature variations, have properties that are similar to those of drift. To reduce all of these errors, we have developed the unconventional Heuristic Drift Reduction (HDR) method. This talk will explain the HDR method and its implementation in both vehicle and personnel tracking applications. Experimental results showing a reduction in heading errors by up to one order of magnitude will also be presented for both applications.

back to top

TerraMax: Team Oshkosh Urban Challenge Robot

John Beck, Oshkosh Corporation 

This presentation will describe the vehicle, overall system architecture, sensors and sensor processing, mission planning system, and the autonomous behavioral controls implemented on TerraMax (TM) for the DARPA Urban Challenge. It will present the performance of some notable autonomous behaviors, experience in implementing these behaviors and results of the Urban Challenge National Qualification Event (NQE) and the Urban Challenge Final Event (UCFE) including lessons learned from experience working with a large robotic truck.

back to top