**
Autonomous Exploration Inc. has been awarded a 2009 NASA Phase I SBIR grant** for the proposal titled Advanced Bayesian Methods for Lunar Surface Navigation. We will be working toward developing navigation systems that target the new Lunar Electric Rover and the Mark III space suits.

**TECHNICAL ABSTRACT**

The key innovation of this project will be the application of advanced Bayesian methods to integrate real-time dense stereo vision and high-speed optical flow with an Inertial Measurement Unit (IMU) to produce a highly accurate planetary rover navigation system. The software developed in this project will leverage current computing technology to implement advanced Visual Odometry (VO) methods that will accurately track much faster rover movements. Our fully Bayesian approach to VO will utilize more information from the images than previous methods are capable of using. Our Bayesian VO does not explicitly select features to track. Instead it implicitly determines what can be learned from each image pixel and weights the information accordingly. This means that our approach can work with images that have no distinct corners, which can be a significant advantage with low contrast images from permanently shadowed areas. We expect that the error characteristics of the visual processing with be complementary to the error characteristics of a low-cost IMU. Therefore, the combination of the two should provide highly accurate navigation.

**POTENTIAL NASA COMMERCIAL APPLICATIONS**

Visual Odometry (VO) has played a key role in Mars exploration with the Spirit and Opportunity Mars Exploration Rovers (MERs). However, limitations in onboard computing power severely limit the speed of movement that can be tracked by MERS VO, requiring an order of magnitude reduction in forward progress in area where VO was required.

The software developed in this project will leverage current computing technology to implement advanced VO methods that will accurately track much faster rover movements. This will greatly increase exploration productivity. This improvement will become even more significant when exploring the more distant planetary bodies.

This project will also investigate whether combining vision with a low-cost, lightweight, low-power Micro-ElectroMechanical System (MEMS) Inertial Measurement Unit (IMU) can produce acceptable accuracy for lunar and planetary exploration. If so, this will facilitate the design of lower-cost, light-weight rovers, which will make it feasible to launch a team of rovers for wide area exploration.

**POTENTIAL NON-NASA COMMERCIAL APPLICATIONS**

There will be many potential terrestrial applications for a Bayesian VO system. Although GPS-IMU systems can work well in open outdoor settings, GPS is degraded or unavailable in indoor settings or in outdoor areas with significant tree cover. A navigation system combining a GPS and an IMU with Bayesian VO could provide continuous operation in all environments. The success of this project should lay the groundwork for low-cost, low-power, light-weight integrated navigation systems for robots and autonomous vehicles operating in a wide range of environments.

One potential market for this technology is the Department of Defense (DoD). Congress has given DoD a mandate that by 2020 30% of ground vehicles should be robotic. An accurate, low-cost VO system should allow many of these vehicles to be semi-autonomous, enabling only supervisory control for many missions.

NASA’s technology taxonomy has been developed by the SBIR-STTR program to disseminate awareness of proposed and awarded R/R&D in the agency. It is a listing of over 100 technologies, sorted into broad categories, of interest to NASA.

TECHNOLOGY TAXONOMY MAPPING |

Guidance, Navigation, and Control
Perception/Sensing |

This session will feature several prominent researchers involved with autonomous or cognitive robotics.

This includes, but is not limited to:

Pierre Bessière, CNRS, Grenoble University

Stephen Roberts, Oxford University, UK

Richard Dearden, University of Birmingham, UK

This special session will focus on the application of Bayesian probability theory to a variety of practical machine learning problems. Bayesian methods provide a unique opportunity to accommodate detailed information about the signal models as well as prior information about problem-specific constraints on parameter values. The recent increase in computational power as well as the development of more powerful sampling techniques have made the application of Bayesian methods to complex problems more practical than ever. This session will focus on a variety of signal processing applications as well as machine intelligence with a focus on robotics, intelligent instrumentation, experimental design, image understanding and vehicle health monitoring. There will be a live demo of a robot employing Bayesian adaptive exploration to solve a search-and-characterize problem.

]]>An abstract follows below:

**Kevin H. Knuth**

Associate Professor of Physics and Informatics, University at Albany

President, Autonomous Exploration, Inc.

In our mutual desire to develop an intuitive understanding of quantum mechanics, John Skilling and I have been working to clarify the foundations of inference.

In this talk I introduce lattice theory as a new foundation for a rational description of the world around us. I will show that fundamental mathematical fields such as measure theory, probability theory and information theory are based on basic symmetries of the lattice and can be easily derived.

Our rational description relies on a partially ordered set (poset) of states. The process of expansion (lattice exponentiation) generates a distributive lattice from a poset. I show that successive expansions applied to the poset of states results in a lattice of all the possible statements that can be made about the system, and a lattice of all the possible questions that can be asked of a system.

Quantification is performed by introducing real-valued functions called valuations. The symmetries of the lattice lead to a unique calculus that obeys a summation property, which is the basis of addition as the fundamental operation in measure theory. Introduction of the notion of context via functions called bi-valuations leads to a product rule for context-based measures. On the lattice of statements, this unique calculus is the probability calculus with its sum and product rule. An additional constraint relating the relevance of a question to the probabilities of the statements that answer it leads to entropy as the basic measure of a question. The unique calculus on the lattice of questions is a new context-based generalization of information theory.

I will finish by introducing a new picture of quantum mechanics where experimental states are partially ordered to form a poset. The valuations on this poset are the quantum amplitudes with Feynman path integrals representing the summation property. Expansion of the state poset to a lattice of statements leads to probabilities whose values depend on the quantum amplitudes. In this way probability theory naturally enters the quantum picture as inference.

]]>The talk involved a demonstration of an autonomous robot that designs its own experiments using Bayesian Adaptive Exploration.

An abstract follows below:

**Kevin H. Knuth**

Associate Professor of Physics and Informatics, University at Albany

President, Autonomous Exploration, Inc.

Most decision making processes must be performed in the face of uncertain or incomplete information. It is critical to be able to make the best inferences one can under these situations, as well as to ask for additional information in an intelligent manner. In this talk, I will introduce a unified framework for automated inference and inquiry based on Bayesian probability theory and Bayesian Adaptive Exploration.

I will begin by considering a system described by a finite set of possible states, and from this I will derive the rules of rational inference by considering the lattice of possible statements that can be made about the system. These rules result in Bayesian probability theory, but seen from a new perspective. I will then introduce Bayesian Adaptive Exploration, which can be used to decide which new data to collect given what is currently known.

I will demonstrate the theory with a Bayesian machine (robot) that locates and characterizes shapes on a playing field. The intelligent aspects of this machine rely on two systems: an inference engine and an inquiry engine. The inference engine simultaneously performs Bayesian model estimation and Bayesian parameter estimation by relying on the nested sampling algorithm. Meanwhile the inquiry engine performs experimental design by identifying the most relevant measurement given the current state of knowledge of the machine. This is accomplished by quantifying the entropy of the possible experimental outcomes. This research represents a real-time implementation of a unified framework of inference and inquiry.

** **

An abstract follows below:

**Kevin H. Knuth**

Associate Professor of Physics and Informatics, University at Albany

President, Autonomous Exploration, Inc.

**Julian Center**

CEO, Autonomous Exploration, Inc.

With an increasing reliance on robotic platforms to perform scientific exploration in remote or hostile environments, it is becoming crucial that robotic systems be able to perform autonomous intelligent sensor placement as well as autonomous experimental design. Such a system requires encoding of scientific knowledge, the ability to make inferences from data, and the ability to identify the most relevant question to ask given both the instrument’s prior knowledge and the issue it is designed to address. This requires implementation of two computational engines: the inference engine and the inquiry engine. Here we demonstrate our first efforts to develop intelligent instruments that rely on autonomous sensor placement.

]]>Kevin Knuth has a laboratory in the physics department of the University at Albany that is filled with LEGOs. The bricks are relatively cheap and can be used to rapidly prototype a robot’s body. Knuth’s robots are being programmed to solve such problems as mapping complex terrain.

At UAlbany Day on Saturday, Oct. 25, he will give a demonstration on

Robotics and Robotic Explorationin Life Sciences Room 143 at 10:45 a.m.

More here:

http://www.albany.edu/news/update_4522.shtml

Building instructions for the robot shown in the UAlbany article can be found on Brickengineer.com

Visit Robots Everywhere for a general blog on robotics news.

]]>The abstract follows:

**Intelligent Science Platforms**

Kevin H. Knuth

Departments of Physics and Informatics, University at Albany, Albany NY 12222

Autonomous Exploration Inc., Albany NY 12208

The exploration of space requires that we continue our dependence on remote scientific platforms. The continued success of the Mars Exploration Rovers highlights the great benefits of navigational autonomy. However, science operations continue to require a team of scientists to select the specific experiments to perform and to precisely guide sensor placement. While this works well on Mars, which has a communication delay ranging from 6.5 to 44 minutes, this model will be strained during future operations on Jupiter’s icy moons, and will most certainly break on future Saturnian missions to Titan and Enceladus. For robotic missions to operate in the outer solar system at a production level comparable to that of the Mars rovers, they will require greater autonomy, not only in mapping and navigation, but also in experimental design and sensor placement.

I will introduce our initial efforts to develop a software-based inquiry engine that relies on a generalized form of information theory called the inquiry calculus. This computational technology enables one to compute the optimal experimental question to ask in a given situation. This technology depends on predicting the probable answers to questions, and selecting the question based on the entropy of the probability distribution of potential answers. I will demonstrate these concepts on a robotic arm that has been programmed to identify and characterize shapes on a playing field using only a simple light sensor.

]]>Kevin Knuth

]]>This lets you write to cheap storage — $8 for a 1GB USB Flash stick these days — and access those same files from Windows or Linux from the filesystem level. Cheaper devices, for instance Futurlec’s SD/MMC Mini Board ($7), let you read and write Flash memory in SD card form, but they require you to (a) speak the SD card language and (b) implement the FAT filesystem access yourself. (Note: C libraries are available for both of these purposes.)

Since the USB mass storage language is standardized, you could probably connect a USB hard disk enclosure to the Memory Stick Datalogger to gain access to very large data storage areas. The product manual is not clear on this point, however, and I haven’t had a chance to test compatibility with hard disks.

]]>