Gordon Vidaver

Evolution-Based Deliberative Planning for Cooperating Unmanned Ground Vehicles in a Dynamic Environment

Many challenges remain in the development of tactical planning systems that will enable automated, cooperative replanning of routes and mission assignments for multiple unmanned ground vehicles (UGVs) under changing environmental and tactical conditions. We have developed such a planning system that uses an evolutionary algorithm to assign waypoints and mission goals to multiple UGVs so that they jointly achieve a set of mission goals. Our evolutionary system applies domain-specific genetic operators, termed tactical advocates because they capture specific tactical behaviors, to make targeted improvements to plans. The plans are evaluated using a set of tactical critics that together comprise a multiobjective fitness function. Each critic evaluates a plan against criteria such as avoiding an enemy or meeting mission goals. Experimental results show that this approach produces highquality plans with the potential for real-time dynamic replanning.

A multiagent society for military transportation scheduling

We are in the process of building a proof-of-concept automated system for scheduling all the transportation for the United States military down to a low level of detail. This is a huge problem currently handled by many hundreds of people across a large number and variety of organizations. Our approach is to use a multiagent society, with each agent performing a particular role for a particular organization. Use of a common multiagent infrastructure allows easy communication between agents, both within the transportation society and with external agents generating transportation requirements. We have demonstrated the feasi-bility of this approach on several large-scale deployment scenarios. Copyright

Scheduling and route selection for military land moves using genetic algorithms

We investigate the problem of scheduling the move of a large amount of military equipment from a fort or depot to a port. This problem differs from traditional distribution scheduling problems in a number of ways including: (i) the trucks need to be grouped into convoys, (ii) there is a single source location and a single destination, and (iii) there are potentially so many trucks traveling the same set of roads that the effects on other traffic must be considered. We have divided the problem into two parts: (i) selecting a fixed set of routes and (ii) forming the trucks into convoys and selecting routes and departure times for each convoy. We describe how we have used genetic algorithms to solve each of these problems. We emphasize how the ability to incorporate domain knowledge into the genetic algorithms has allowed us to easily create algorithms well suited to the particular constraints of the problems.

Flexible and Purposeful NPC Behaviors using Real-Time Genetic Control

There is an increasing need in modern computer games for non-player characters (NPCs) with robust behaviors that achieve game objectives while appearing flexible and believable to the human players interacting with those NPCs. Evolutionary approaches to game artificial intelligence (game AI) have produced successful results for complex game winning strategies, realistic behavior patterns for groups of simulated entities, and more. However, there has been relatively little effort on evolutionary techniques for producing rich NPC behaviors for interaction with human players. To explore whether evolutionary mechanisms can support real-time control of NPCs to produce flexible and purposeful behavior, we present our initial efforts at integrating a genetic algorithm based robotic controller with an off-the-shelf game to control one or more NPCs dynamically. We describe the integration effort and our initial observations, and discuss our plan for achieving richer NPC control and for performing more detailed analysis of the behaviors of the resulting NPCs.

Automated hardware design using genetic programming, VHDL, and FPGAs

We have developed a completely automated approach to hardware design based on integrating three core technologies into one comprehensive system, namely genetic programming (GP), the VHSIC Hardware Description Language (VHDL) and field programmable gate arrays (FPGAs). Our system uses an automated GP engine, as opposed to a human designer, to evolve a hardware design composed of one or more FPGAs that will maximally achieve an applications software requirements. Several variants of our system exist; other variants are currently under development. The focus of this paper is to describe our original system design and its most recent revision to date.

A reconfigurable multiagent society for transportation scheduling and dynamic rescheduling

We have investigated the use of an agent-based system to automate the process of scheduling all the transportation assets for the U.S. military. For this problem, the creation of a fixed schedule based on static requirements is insufficient. Instead, we need to be able to maintain the schedule as a dynamic entity in the face of changing requirements, unreliable assets, and unexpected events during execution. Towards this end, we have focused on two main areas: dynamic rescheduling and dynamic load management. Dynamic rescheduling involves updating an existing schedule in response to changes in the scheduling resources and requirements, and is a special challenge in a multiagent environment. We discuss the approach we used to propagate and coordinate changes between the different agents. Dynamic load management includes the ability of the multiagent society to change its structure in response to changing computational requirements. We examine a set of techniques, including dynamic agent creation, to continually update the allocation of computational resources to the scheduling process.

A Genetic-Algorithm-Based Reconfigurable Scheduler

Scheduling problems vary widely in the nature of their constraints and optimization criteria. Most scheduling algorithms make restrictive assumptions about the constraints and criteria and hence are applicable to only a limited set of scheduling problems. A reconfigurable scheduler is one that, unlike most schedulers, is easily configured to handle a wide variety of scheduling problems with different types of constraints and criteria. We have implemented a reconfigurable scheduler, called Vishnu, that handles an especially large range of scheduling problems. Vishnu is based upon a genetic algorithm that feeds task orderings to a greedy scheduler, which in turn allocates those tasks to a schedule. The scheduling logic (i.e. constraints and optimization criteria) is reconfigurable, and Vishnu includes a general and easily expandable framework for expressing this logic using hooks and formulas. The scheduler can find an optimized schedule for any problem specified in this framework. We illustrate Vishnu’s flexibility and evaluate its performance using a variety of scheduling problems, including some classic ones and others from real-world scheduling projects.

Tactical UGV navigation and logistics planning

The Armys push towards developing highly flexible military teams that combine manned and unmanned units requires significant advances in the intelligence of the unmanned units and in the tools used to provide logistical support. BBN Technologies has recently completed a simulation-based project for the Army Research Lab in which we applied an evolutionary computation approach for determining the tactical responses of an unmanned ground vehicle, moderated by explicit rules of engagement. BBN is also currently developing a logistic analysis prototype that uses agent-based technology and evolutionary computation to enable rapid logistics planning and replanning for supporting an Army organization encountering a diversity of planned and unplanned situations.