Scott D. Anderson

Simulation of multiple time-pressured agents

The paper describes a simulation substrate that allows thinking agents to interact with a world. The world is simulated by standard discrete event simulation, but the timing of an agent’s behavior is determined by the amount of computation it performs. Therefore, if an agent thinks a lot about what to do given a situation in the world, the duration of its thinking results in a delay to its subsequent actions. Thus, the thinking of the agent is time pressured. The computation time of the agent is automatically assessed by the substrate in a way that is independent of the computer running the simulation. This is done by implementing the thinking of the agents in a variant of Common Lisp called Timed Common Lisp, in which each function advances a clock by an appropriate, user-specifiable amount of time. This renders agent thinking and behaviors deterministic, making results comparable and replicatable across platforms. The simulation substrate also supports the interaction of continuous activities, in addition to executing discrete, point-like events. This substrate has been used to implement an Artificial Intelligence Planning system that simulates multiple agents fighting forest fires in Yellowstone National Park.

A TOOLBOX FOR ANALYZING PROGRAMS

As program behavior becomes complex, it’s increasingly important to analyze their behavior statistically. The article describes two separate but synergistic tools for statistically analyzing large Lisp programs. The first tool, called CLIP (Common Lisp Instrumentation Package), allows the researcher to define and run experiments, including experimental conditions (parameter values of the planner or simulator) and data to be collected. The data are written out to data files that can be analyzed by statistics software. The second tool, called CLASP (Common Lisp Analytical Statistics Package), allows the researcher to analyze data from experiments by using graphics, statistical tests, and various kinds of data manipulation. CLASP has a graphical user interface (using CLIM, the Common Lisp Interface Manager) and also allows data to be directly processed by Lisp functions. Finally, the paper describes a number of other data-analysis modules that have been added to work with CLIP and CLASP.

Timed Common Lisp: the duration of deliberation

The two key features of Deliberation Scheduling and Anytime Algorithms are the duration of the computation and the resulting quality. Clearly, quality can be difficult to define and highly dependent on the domain. Duration, on the other hand, seems straightforward: how long the computation takes. But on what processor? Should the processor matter? What code counts in the computation? How is that codes duration modeled? These questions are addressed in this paper.

Understanding robot navigation and control

Robot navigation and control are means by which robots can move around in their environment to accomplish tasks. This extended abstract reports on some research that we are currently conducting on control strategies for robot navigation. In particular, comparisons between sonar navigation and dead reckoning were made, to see to what extent the robot could use sonar information to increase the accuracy of its movements.

Tools for experiments in planning

The paper describes two separate but synergistic tools for running experiments on large Lisp systems such as artificial intelligence planning systems, by which we mean systems that produce plans and execute them in some kind of simulator. The first tool, called CLIP (Common Lisp Instrumentation Package), allows the researcher to define and run experiments, including experimental conditions (parameter values of the planner or simulator) and data to be collected. The data are written out to data files that can be analyzed by statistics software. The second tool, called CLASP (Common Lisp Analytical Statistics Package), allows the researcher to analyze data from experiments by using graphics, statistical tests, and various kinds of data manipulation. CLASP has a graphical user interface (using CLIM, the Common Lisp Manager) and also allows data to be directly processed by Lisp functions.<<ETX>>

Two ways to act

We derive a simple mathematical model of the interaction of two tasks running concurrently on a uniprocessor. Specifically, we model the rate at which two periodic tasks will interrupt each other. The predictions of this model are supported by simulation experiments. From the model, we argue that placing reflexive control and cognitive control on different processors is not justified by the different time-scales over which they act, but is justified by the extent to which the higher priority task dominates the uniprocessor.