Publications Articles with Category: Planning and Control


Plotting a Better Path

July 3, 2012
Stilwell D, Plotting a Better Path, VaCAS, July 3 2012

What is the easiest way to determine what is on the ocean floor? How quickly can robot explorers tell the difference between a sand ripple, hidden treasure, or even a possible danger? ECE associate professor Dan Stilwell’s team in the Autonomous Systems and Controls Laboratory (ASCL) is developing answers to these questions.

With a new three-year grant fromthe Office of Naval Research, they are working on the algorithms to deploy autonomous underwater vehicle (AUV) and sensing technology in the most efficient manner. “If we have assets that can characterize the environment, how would we deploy those assets?” Stilwell asks.

Mapping everything, he says, takes a long time and uses more power and resources than necessary, so his team wishes to efficiently search for underwater objects with a given probability that every object of interest has been identified.


Navigating blindfolded

November 8, 2011
Stauffer N, Navigating blindfolded, MIT News, Nov 2011

Advanced mathematical techniques enable AUVs to survey large, complex and cluttered seascapes.


Intelligent Planning and Assimilation of AUV – Obtained Measurements within a ROMS – Based Ocean Modeling System

August 21, 2011
Davini B, Choboter P, Clark C, Intelligent Planning and Assimillation of AUV – Obtained Measurements within a ROMS – Based Ocvean Modeling System, UUST, Aug 21 2011

An efficient method is presented for collecting oceanic data with autonomous underwater vehicles (AUVs) and assimilating that data into a numerical ocean model. A system based on the data assimilation tools of the Regional Ocean Modeling System (ROMS) is developed that intelligently plans for and integrates AUV measurements with the goal of minimizing model standard deviation. An algorithm for selecting AUV paths is described that seeks to improve the model accuracy by gathering data in high-interest locations. This algorithm and its e ect on the ocean model accuracy are tested by comparing the results of missions made with the algorithm (i.e. optimial) with missions that follow a standard lawn-mower pattern (i.e. non-optimal). The results of the experiments demonstrate that the system is successful in improving the ROMS ocean model accuracy. Also shown are results comparing optimized missions and non-optimized missions.


Autonomous Robots in the Fog of War

August 2, 2011
Weiss L, Robots in the Fog of War, IEEE Spectrum, Aug 2011

Networks of autonomous robots will someday transform warfare, but significant hurdles remain

Why haven’t we seen a fully autonomous robot that can sense for itself, decide for itself, and seamlessly interact with people and other machines? Unmanned systems still fall short in three key areas: sensing, testing, and interoperability. Although the most advanced robots these days may gather data from an expansive array of cameras, microphones, and other sensors, they lack the ability to process all that information in real time and then intelligently act on the results. Likewise, testing poses a problem, because there is no accepted way to subject an autonomous system to every conceivable situation it might encounter in the real world. And interoperability becomes an issue when robots of different types must interact; even more difficult is getting manned and unmanned systems to interact.


Huxley: A Flexible Robot Control Architecture for Autonomous Underwater Vehicles

June 15, 2011
Goldberg D, Huxley: A Flexible Robot Control Architecture for Autonomous Underwater Vehicles, IEEE Oceans 2011, June 2011


This paper presents “Huxley,” a production robot control architecture that was developed by Bluefin Robotics for its fleet of autonomous underwater vehicles (AUVs). Huxley was designed with flexibility foremost in mind, allowing it to be easily adapted and reliably deployed on a wide range of platforms. The architecture follows a layered paradigm, providing a clean and logical abstraction for the major control functions. It also provides an interface for interaction with the layers, enabling expansion of the core capabilities of the architecture. This interface provides users the flexibility to develop smart payloads capable of utilizing available data, modifying the behavior of the AUV and even exploring new frontiers of autonomy.


Exploring Trade-offs in AUV Controller Design for Shark Tracking

March 25, 2011
Bertsch Louis J. Exploring Trade-offs in AUV Controller Design for Shark Tracking, Cal Poly University, March 2011

This thesis explores the use of an Autonomous Underwater Vehicle (AUV) to track and pursue a tagged shark through the water. A controller was designed to take bearing and range to the shark tag and then control the AUV to pursue it.

First, the ability of a particle lter to provide an accurate estimation of the location of the shark relative to the AUV is explored. Second, the ability of the AUV to follow the shark0s path through the water is shown. This ability allows for localized environmental sampling of the shark0s preferred path. Third, various path weightings are used to optimize the efficiency of pursuing the shark. This demonstrates that the proposed controller is efficient and effective. Fourth, the bene ts of the addition of a second AUV are explored and quanti ed. The secondary AUV is shown to maintain formation without direct communication from the primary AUV. However, the communication of the AUVs increases the accuracy of all measurements and allows for future expansion in the complexity of the controller. Fifth, the e ects of predicting the shark’s future movement is explored. Sixth, the e ect of noise in the signal from the shark tag is tested and the level of noise at which the AUV can no longer pursue the shark is shown. This investigates the real world ability of the controller to accept noisy inputs and still generate the appropriate response. Finally, the positive results of the previous sections are combined and tested for various noise levels to show the improved controller response even under increased noise levels.

To validate the proposed estimator and controller, seven tests were conducted. All tests were conducted on existing shark path data recorded by a stationary acoustic receiver and a boat mounted acoustic receiver. Tests were conducted on data sets from two di erent species of sharks, (Shovelnose and White) with two very di erent swimming behaviors. This shows the solution”s fexibility in the species of shark tracked.



August 24, 2010

The Joint Architecture for Unmanned Systems (JAUS) is an international standard of the SAE AS-4 Unmanned Systems Steering Committee. The OpenJAUS project team has recently undergone an effort to update their software to support the new SAE JAUS standards (AS5684, AS5669 and AS5710). This paper will discuss the critical design issues that needed to be overcome in order to develop an effective JAUS solution. Also, several features that were added to the Open-JAUS design to address advanced requirements currently outside of the scope of JAUS will be discussed. The paper will highlight how the OpenJAUS team conducted their development work in a distributed online manner, and give examples of the tools used for this process. An approach will be given for how the new OpenJAUS features will be offered to AS-4 for future standardization.


An Overview of Autonomous Underwater Vehicle Research and Testbed at PeRL

April 30, 2009
Brown Hunter C, Kim Ayoung, Eustice Ryan M., An Overview of Autonomous Underwater Vehicle Research and Testbed at PeRL, MTS 43-2 Spring 2009

This article provides a general overview of the autonomous underwater vehicle (AUV) research thrusts being pursued within the Perceptual Robotics Laboratory (PeRL) at the University of Michigan. Founded in 2007, PeRL’s research centers on improving AUV autonomy via algorithmic advancements in environmentally based perceptual feedback for real-time mapping, navigation, and control.

Our three major research areas are (1) real-time visual simultaneous localization and mapping (SLAM), (2) cooperative multi-vehicle navigation, and (3) perceptiondriven control. Pursuant to these research objectives, PeRL has developed a new multi-AUV SLAM testbed based upon a modified Ocean-Server Iver2 AUV platform.

PeRL upgraded the vehicles with additional navigation and perceptual sensors for underwater SLAM research. In this article, we detail our testbed development, provide an overview of our major research thrusts, and put into context how our modified AUV testbed enables experimental real-world validation of these algorithms.

Keywords: AUVs, SLAM, navigation, mapping, testbed


Cooperative Mapping and Navigation for Multiple Unmanned Underwater Vehicles

April 24, 2000
Leonard J, Cooperative Mapping and Navigation for Multiple Unmanned Underwater Vehicles, International UUV Symposium Apr 24 2008, Newport RI


This Paper considers the problem of cooperative mapping and navigation (CMAN) by multiple unmanned underwater vehicles (UUVs). The goal is for several UUVs to concurrently build maps of an unknown environment, and to use these maps for navigation. This work builds on our previous research in development of concurrent mapping and localization (CML) techniques for a single vehicle. In this paper, cooperative stochastic mapping is proposed as a new framework for featurebased CML by multiple vehicles. Previous research related to cooperative mapping and navigation is reviewed. New research issues encountered, such as information transfer management, decentralized data fusion, and cooperative adaptive sampling are discussed.