Jacob Beal

TAAS Journal 2018 Journal Article

Adaptive Opportunistic Airborne Sensor Sharing

• Jacob Beal
• Kyle Usbeck
• Joseph Loyall
• Mason Rowe
• James Metzler

Airborne sensor platforms are becoming increasingly significant for both civilian and military operations; yet, at present, their sensors are typically idle for much of their flight time, e.g., while the sensor-equipped platform is in transit to and from the locations of sensing tasks. The sensing needs of many other potential information consumers might thus be served by sharing such sensors, thereby allowing other information consumers to opportunistically task them during their otherwise unscheduled time, as well as enabling other improvements, such as decreasing the number of platforms needed to achieve a goal and increasing the resilience of sensor tasks through duplication. We have implemented a prototype system realizing these goals in Mission-Driven Tasking of Information Producers (MTIP), which leverages an agent-based representation of tasks and sensors to enable fast, effective, and adaptive opportunistic sharing of airborne sensors. Using a simulated large-scale disaster-response scenario populated with publicly available Geographic Information System (GIS) datasets, we demonstrate that correlations in task location are likely to lead to a high degree of potential for sensor-sharing. We then validate that our implementation of MTIP can successfully carry out such sharing, showing that it increases the number of sensor tasks served, reduces the number of platforms required to serve a given set of sensor tasks, and adapts well to radical changes in flight path.

TAAS Journal 2017 Journal Article

Self-Adaptation to Device Distribution in the Internet of Things

• Jacob Beal
• Mirko Viroli
• Danilo Pianini
• Ferruccio Damiani

A key problem when coordinating the behaviour of spatially situated networks, like those typically found in the Internet of Things (IoT), is adaptation to changes impacting network topology, density, and heterogeneity. Computational goals for such systems, however, are often dependent on geometric properties of the continuous environment in which the devices are situated rather than the particulars of how devices happen to be distributed through it. In this article, we identify a new property of distributed algorithms, eventual consistency, which guarantees that computation converges to a final state that approximates a predictable limit, based on the continuous environment, as the density and speed of devices increases. We then identify a large class of programs that are eventually consistent, building on prior results on the field calculus computational model (Beal et al. 2015; Viroli et al. 2015a) that identify a class of self-stabilizing programs. Finally, we confirm through simulation of IoT application scenarios that eventually consistent programs from this class can provide resilient behavior where programs that are only converging fail badly.

KER Journal 2016 Journal Article

Trading accuracy for speed in approximate consensus

• Jacob Beal

Abstract Approximate consensus is an important building block for distributed systems, used overtly or implicitly in applications as diverse as formation control, sensor fusion, and synchronization. Laplacian-based consensus, the current dominant approach, is extremely accurate and resilient, but converges slowly. Comparing Laplacian-based consensus to exact consensus algorithms, relaxing the requirements for accuracy and resilience should enable a spectrum of algorithms that incrementally tradeoff accuracy and/or resilience for speed. This manuscript demonstrates that may be so by beginning to populate this spectrum with a new approach to approximate consensus, Power-Law-Driven Consensus (PLD-consensus), which accelerates consensus by sending values across long distances using a self-organizing overlay network. Both a unidirectional and bidirectional algorithm based on this approach are studied. Although both have the same asymptotic O ( diameter ) convergence time (vs. O ( diameter 2 ) for Laplacian-based), unidirectional PLD-consensus is faster and more resilient than bidirectional PLD-consensus, but exhibits higher variance in the converged value.

TAAS Journal 2015 Journal Article

Superdiffusive Dispersion and Mixing of Swarms

• Jacob Beal

A common swarm task is to disperse evenly through an environment from an initial tightly packed formation. Due to communication and sensing limitations, it is often necessary to execute this task with little or no communication between swarm members. Unfortunately, prior approaches based on repulsive forces or uniform random walks can often converge quite slowly. With an appropriate choice of random distribution, however, it is possible to generate optimal or near-optimal dispersion and mixing in swarms with zero communication. In particular, we discuss three extremely simple algorithms: reactive Levy walk, reactive ball dispersion, and purely reactive dispersion. All three algorithms vastly outperform prior approaches in both constrained and unconstrained environments, providing a range of options for trading off between aggressiveness and evenness in dispersion.

AAAI Conference 2013 Conference Paper

A Morphogenetically Assisted Design Variation Tool

• Aaron Adler
• Fusun Yaman
• Jacob Beal
• Jeffrey Cleveland
• Hala Mostafa
• Annan Mozeika

The complexity and tight integration of electromechanical systems often makes them “brittle” and hard to modify in response to changing requirements. We aim to remedy this by capturing expert knowledge as functional blueprints, an idea inspired by regulatory processes that occur in natural morphogenesis. We then apply this knowledge in an intelligent design variation tool. When a user modifies a design, our tool uses functional blueprints to modify other components in response, thereby maintaining integration and reducing the need for costly search or constraint solving. In this paper, we refine the functional blueprint concept and discuss practical issues in applying it to electromechanical systems. We then validate our approach with a case study applying our prototype tool to create variants of a miniDroid robot and by empirical evaluation of convergence dynamics of networks of functional blueprints.

TAAS Journal 2011 Journal Article

Spatial Computing

• Jacob Beal
• Olivier Michel
• Ulrik Pagh Schultz

AAMAS Conference 2010 Conference Paper

Laplacian-Based Consensus on Spatial Computers

• Nelson Elhage
• Jacob Beal

Robotic swarms, like all spatial computers, are a challenging environment for the execution of distributed consensus algorithms due to their scale, diameter, and frequent failures. Exact consensus is generally impractical on spatial computers, so we consider approximate consensus algorithms. In this paper, we show that the family of self-organizing protocols based on the graph Laplacian of a network are impractical as well. With respect to the structure of a finite-neighborhood spatial computer, we find that these protocols have an expected convergence time of $O(diameter^2)$ when the inputs are strongly correlated with location. Verifying this result in simulation, we further determine that the constant factor on the convergence time is high, rendering Laplacian-based approximate consensus unsuitable for general use on spatial computers.

IJCAI Conference 2009 Conference Paper

• Jacob Beal

Self-organizing maps can be used to implement an associative memory for an intelligent system that dynamically learns about new high-level domains over time. SOMs are an attractive option for implementing associative memory: they are fast, easily parallelized, and digest a stream of incoming data into a topographically organized collection of models where more frequent classes of data are represented by higher-resolution collections of models. Typically, the distribution of models in an SOM, once developed, remains fairly stable, but developing expertise in a new high-level domain requires altering the allocation of models. We use a mixture of analysis and empirical studies to characterize the behavior of SOMs for high-level associative memory, finding that new high-resolution collections of models develop quickly. High-resolution areas of the SOM decay rapidly unless actively refreshed, but in a large SOM, the ratio between growth rate and decay rate may be high enough to support both fast learning and long-term memory.

IS Journal 2009 Journal Article

Guest Editors' Introduction: The New Frontier of Human-Level Artificial Intelligence

• Jacob Beal
• Patrick H. Winston

Within the field of human-level intelligence, researchers are combining a variety of approaches toward the goals of human-like breadth, flexibility, and resilience for artificial intelligence systems. Each of the four papers in this special issue brings a different background and perspective on the subject, and hence a different technical approach.

AAMAS Conference 2008 Conference Paper

Shared Focus of Attention for Heterogeneous Agents

• Jacob Beal

A network of cooperating agents must be able to reach rough consensus on a set of topics for cooperation. With highly heterogeneous agents, however, incommensurable measures and imprecise translation render ordinary consensus algorithms inappropriate. I present a distributed mechanism for shared focus of attention that begins to address these problems, using an engineered emergence approach inspired by recent results on the dynamics of evolution in systems with spatial extent. Simulation shows that the algorithm converges in time proportional to the diameter of the network and gives a range of reasonable settings for the parameters.

AAAI Conference 2005 Short Paper

Leveraging Language into Learning

• Jacob Beal

I hypothesize that learning a vocabulary to communicate between components of a system is equivalent to general learning. Moreover, I assert that some problems of general learning, such as eliminating bad hypotheses, deepening shallow representations, and generation of near-misses, will become simpler when refactored into communication learning problems.