David Scheidt

Design of a new independently-mobile reconfigurable modular robot

A new self-reconfigurable robot is presented. The robot is a hybrid chain/lattice design with several novel features. An active mechanical docking mechanism provides inter-module connection, along with optical and electrical interface. The docking mechanisms function additionally as driven wheels. Internal slip rings provide unlimited rotary motion to the wheels, allowing the modules to move independently by driving on flat surfaces, or in assemblies negotiating more complex terrain. Modules in the system are mechanically homogeneous, with three identical docking mechanisms within a module. Each mechanical dock is driven by a high torque actuator to enable movement of large segments within a multi-module structure, as well as low-speed driving. Preliminary experimental results demonstrate locomotion, mechanical docking, and lifting of a single module.

Cooperating Unmanned Vehicles

Effective employment of autonomous multi-vehicle control can meet a critical need in todays military. Swarming unmanned ground and aerial vehicles can achieve militarily effective cooperative action autonomously, using reactive, effects-based behaviors and broadcast-only, decentralized communications. We achieve this control through stigmergic potential fields, a form of cooperative motor schema behavior based upon mathematical functions that are associated with entities in the operational environment of the vehicle. We have demonstrated empirically, as well as through hardware-in-the-loop testing, how reactive swarming behaviors using stigmergic potential fields can offer robustly sufficient behavior with improved total system survivability and total operational effectiveness, particularly under dynamic environmental and operational conditions.

Cooperative Localization and Mapping

Practical realization of autonomous multi-vehicle control in feature rich GPS-denied areas requires that agents possess the capability to generate and share maps with details adequate for successful operation in complex environments. In addition, coordination of behaviors among robotic vehicles in a swarm is greatly enhanced by exploitation of information acquired by all members of the group. The representation of, and mechanism for this information exchange and the control algorithms that utilize this information are the key components of autonomous multi-vehicle control. The use of stigmergic potential fields (SPF) and distributed control algorithms to realize adaptive, cooperative behaviors within the context of feature-rich GPS-denied environments is the focus of this paper. Specifically, the described control algorithms have been combined with the leading simultaneous localization and mapping (SLAM) approaches to yield robust multi-robot behaviors including exploration and mapping

Wireless Network Communications Architecture for Swarms of Small UAVs

Swarms of small, autonomous, unmanned aerial vehicles (UAVs) are true force multipliers, enabling soldiers in the field a birds-eye view of their environment, and providing a real-time early warning system for convoys, ground troops, and air assets. These next generation multi-vehicle UAV systems work as a collaborative autonomous unit, receiving only high-level mission commands, and require little human intervention for control. The Johns Hopkins University Applied Physics Laboratory (JHU/APL) is involved in a multi-year R&D program to develop “strong autonomy” based on innovative scalable architectures for high-level autonomy and emergent group behavior in distributed vehicle systems that include both UAVs and small robots. JHU/APL has developed a reliable, robust communications architecture to manage the input and output of “belief” messages. This architecture comprises the content, structure, prioritization, broadcast scheduling, and reception of these messages. The architecture also includes the physical wireless link between vehicles, which for these proof-of-concept demonstrations is an ad hoc IEEE 802.11b wireless local area network (WLAN). The communications architecture supports three primary services: Periodic Packet Multicast service, On-Demand Packet Broadcast service, and Operation, Administration, and Management (OAM) service. The former two services are for transport of beliefs. The OAM service allows configuration of internal default parameters and allows access to its internal status and statistical counters. The architecture provides these services by supporting three interfaces: the Belief Transmit Service Interface (BTSI), Belief Receive Service Interface (BRSI), and OAM Interface (OAMI). Along with the architecture description, this paper presents actual field demonstration results of multiple ground and aerial vehicle systems.

A Control System Test Bed for Demonstration of Distributed Computational Intelligence Applied to Reconfiguring Heterogeneous Systems

This paper presents a combined hardware/software simulation framework for testing and evaluating decentralized control methodologies for integrated mechanical and electrical systems on board US Navy ships. Integrated ship system control is a challenging domain as it involves a set of highly complex and inter-dependent systems, including electrical power, HVAC, and an underlying communications network. This control problem is especially difficult when damage events or current operational objectives require collaboration between the individual systems to make intelligent tradeoffs that maximize overall effectiveness.

Reconfiguring connected resource distribution systems

In this paper we consider complex mechanical systems that operate within harsh, uncertain environments where mission success depends critically upon the systems ability to monitor and maintain its own internal health. In particular, we focus on the problem of reconfiguring a systems internal resource distribution subsystems in response to one or more damage events. We present a graph-based approach that guarantees the resulting states adhere to a desirable resource segregation constraint. We also provide results using this approach in conjunction with optimization techniques to reconfigure a representative hardware test platform.

Demonstration of Effects-Based Operations Using Fully Autonomous Heterogeneous Vehicle Swarms

Effects-based operations have been demonstrated on hardware to realize practical and autonomously performed behaviors utilizing a variety of vehicle platforms including ground, water surface, and aerial vehicles. The heterogeneous vehicle swarms enact a variety of cooperative behaviors autonomously, using reactive, effects-based algorithms and broadcast-only, decentralized communications. Control is achieved through the use of stigmergic potential fields, a form of cooperative motor schema behavior based upon mathematical functions that are associated with entities in the operational environment of the vehicle. The video demonstration shows footage from actual hardware tests that have demonstrated empirically how reactive swarming behaviors using stigmergic potential fields can offer robustly sufficient behavior with improved total system survivability and total operational effectiveness, particularly under dynamic environmental and operational conditions.

Integrating Multi-Agent Systems: A Case Study

Intelligent Agent technology provides an appropriate framework to integrate knowledge with efficient production actions in distributed organizations. Integration of knowledge depends on balanced information representation within and across heterogeneous organizations. Integrating information within a specific environment can be helped by the deployment of standards and common practices. However, it is harder to attempt such a smooth integration with the information of foreign organizations. It is the challenge of this paper to present an architecture that provides a first step towards successful integration of separate multi-agent systems in a real life control domain.

Reusability of knowledge for deriving latent situational information in EW scenarios

this paper provides an overview of research to apply human-machine Situation Awareness (SA) principles to Electronic Warfare (EW). The research leverages the Endsley SA model combined with Cognitive Radio Ontology extensions to provide automatic inference capabilities for EW. The result is an interoperable language for autonomous cognitive EW systems to create and share knowledge about the Electromagnetic Spectrum (EMS) and the platforms using it. In this paper, we explore several modeling and inference constructs using ontologies expressed in the Web Ontology Language (OWL). We show that these constructs are aligned with the elements of the Endsley model and when implemented in an autonomous system they can significantly enhance the SA of its users and lead to more informed courses of action.

A column generation approach for optimized routing and coordination of a UAV fleet

Unmanned Aerial Vecicles (UAVs) in civil and military applications are becoming increasingly popular. Various platform types have already shown their great potential in missions that require rapid surveillance capabilities or logistic support. Large scale incidents require the deployment of several platforms with various capabilities. In this case, coordinated use will lead to more efficient use of the given resources. Problems to resolve resemble known optimization problems from the field of vehicle routing or scheduling. The problem considered in this work includes a given team of homogenous UAVs and a set of target locations with certain requests that need to be served. It is modeled as a variant of the Vehicle Routing Problem (VRP) that is known to be NP hard, i.e. until now no algorithm is known that can solve the problem in polynomial run-time. In this paper, the problem is formulated using a path flow formulation and a column generation algorithm has been implemented and tested to solve simulated realtime instances of the problem in suitable time∗.