Project Summary

Experience with robots at disaster sites suggests that useful emergency response robots must have several characteristics. From the practical mechanical point of view, they must possess basic mechanical durability, very high mobility in complex terrains, simple manipulation capabilities, and the ability to recover from errors and/or failures (such as toppling). Furthermore, teams of mobile robots must be able to gather large amount of sensory information, which is then processed and presented to remote human operators in the correct geometrical context. In order to introduce some of the main issues addressed by this project, consider the task of biohazard assessment. A robot team equipped with a heterogeneous suite of sensors (e.g., cameras, range finders, electronic noses, and biochips) must disperse and rapidly search an area to locate, identify, and catalog chemical/biological/radioactive hazards. The team must also make detailed maps of the hazard site, which will serve as the basis for remediation planning. In addition, individual robots need to share collected information and communicate it in a condensed form to remotely sited human operators. Optimally, the team should be able to adaptively allocate its resources on the fly as it encounters evolving situations. All of these operations rely heavily upon coordination and planning techniques that take sensing into account. Based on the Scout and MegaScout robotic platforms (developed at the University of Minnesota's Center for Distributed Robotics), the creation of a set of algorithms that address these issues is undertaken. The multi-disciplinary team consists of researchers from the University of Minnesota (UMN), the California Institute of Technology (Caltech), and the University of Pennsylvania (UPenn). The intellectual merit of the proposed research can be summarized into:

Introduction of distributed algorithms for effective collaborative sensing. The work spans from simple methods for robust determination of visual landmarks through the use of omnidirectional systems to coordinated movement and placement of Scouts and MegaScouts in order to maximize the quality and relevance of the retrieved information.
Development of distributed localization/mapping methods that leverages the multiple capabilities that heterogeneous teams of robots possess.
In-depth study of the communication issues and emphasis on the transparent integration of ultra wideband communication methodologies to address the communication challenges present in multi-robot systems.
Development of methods for reliable team coordination.
Creation of algorithms for effective information dissemination to users.
Experimental validation of the scalability of the aforementioned algorithms and techniques. This work will provide a comprehensive performance analysis for robotic teams of increasing size in the context of coordinated exploration and information gathering.
Consultation with public agencies (e.g., FEMA) and collaboration with privately owned companies. This process in effect will instigate further interest and lay the foundation for bridging the gap between multi-robot academic research and public acceptance of robotic emergency response teams.

Top

Participating Researchers and Laboratories

University of Minnesota	Nikolaos Papanikolopoulos (PI)	(Computer Science)
	Maria Gini	(Computer Science)
	Richard Voyles	(Computer Science)
	Stergios Roumeliotis	(Computer Science)
	Baoquan Chen	(Computer Science)
	Georgios B. Giannakis	(Electrical and Computer Engineering)
California Institute of Technology	Joel Burdick	(Mechanical Engineering)
University of Pennsylvania	Kostas Daniilidis	(Computer and Information Science)
	George J. Pappas	(Electrical and Systems Engineering) & (Computer and Information Science)

Top

Demos

Top

Images

Top

Publications

Top