Developing World Model Data Specifications as Metrics for Sensory Processing for On-Road Driving Tasks

Published

August 18, 2003

Author(s)

Tony Barbera, John A. Horst, Craig I. Schlenoff, Evan K. Wallace, David Aha

Abstract

Building knowledge-intensive real-time ntelligent control systems is one of the most difficult tasks that humans attempt. It is motivated by the desire to create an artificial reasoning system that displays intelligent behavior (i.e. that can act on the world and successfully accomplish activities that are only possible with the levels of knowledge processing exhibited by human beings). Measuring and evaluating the success of such systems is difficult - a system?s observable behavior is not always indicative of its correctness or quality. This is especially true in complex real-time control systems such as autonomous on-road driving, which is the focus of our Defense Advanced Research Project Agency (DARPA) Mobile Autonomous Robot Software (MARS) On- Road Driving Project. We are performing a task analysis and developing performance metrics for autonomous on-road driving, and using the NIST Real-time Control System (RCS, now referred to as 4D/RCS) [1] design methodology and reference architecture to develop a task decomposition representation format for on-road driving task knowledge. This representation is used as the framework to further specify the world model entities, attributes, features, and events required for proper reasoning about each of the subtask activities. These world model specifications, in turn, are used as the requirements for the sensory processing system; they identify objects that have to be measured in the environment, including their resolutions, accuracy tolerances, detection timing, and detection distances for each subtask activity. We describe our project?s task and world modeling knowledge, exemplify their application, and describe a set of performance metrics for validating sensory processing activities by evaluating the world model representations the system produces for each individual component subtask activity. In this way, taxonomies of autonomous capabilities can be developed and tested against these sensory processing and world model building performance metrics.

Proceedings Title

Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop

Conference Location

Gaithersburg, MD, USA

Conference Title

Performance Metrics for Intelligent Systems (PerMIS) Workshop

Pub Type

Conferences

Download Paper

Local Download

Keywords

Control, finite state machines, Knowledge Engineering, On-road driving, performance metrics, sensory processing, task decomposition, Unmanned Systems

Citation

Barbera, T. , Horst, J. , Schlenoff, C. , Wallace, E. and Aha, D. (2003), Developing World Model Data Specifications as Metrics for Sensory Processing for On-Road Driving Tasks, Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop, Gaithersburg, MD, USA, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=822563 (Accessed April 25, 2024)