A quantitative analysis of memory requirement and generalization performance for robotic tasks

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In autonomous agent systems, memory is an important element to handle agent behaviors appropriately. We present the analysis of memory requirements for robotic tasks including wall following and corridor following. The robotic tasks are simulated with sensor modeling and motor actions in noisy environments. In this paper, control structures are based on finite state machines for memory-based controllers, and we use the evolutionary multiobjective optimization approach with two objectives, behavior performance and memory size. For each task, a quantitative approach to estimate internal states with a different number of sensors is applied and the best controllers are evaluated in several test environments to examine their generalization characteristics and efficiency. Finite state machines with a hierarchy of memory are also compared with feedforward neural networks for the behavior performance.

Original languageEnglish
Title of host publicationProceedings of GECCO 2007
Subtitle of host publicationGenetic and Evolutionary Computation Conference
Pages285-292
Number of pages8
DOIs
Publication statusPublished - 2007 Aug 28
Event9th Annual Genetic and Evolutionary Computation Conference, GECCO 2007 - London, United Kingdom
Duration: 2007 Jul 72007 Jul 11

Publication series

NameProceedings of GECCO 2007: Genetic and Evolutionary Computation Conference

Other

Other9th Annual Genetic and Evolutionary Computation Conference, GECCO 2007
CountryUnited Kingdom
CityLondon
Period07/7/707/7/11

Fingerprint

Quantitative Analysis
Robotics
Data storage equipment
Requirements
Chemical analysis
Finite automata
State Machine
Controller
Sensor
Evolutionary multiobjective Optimization
Controllers
Autonomous agents
Agent Systems
Autonomous Agents
Feedforward neural networks
Sensors
Feedforward Neural Networks
Autonomous Systems
Multiobjective optimization
Chemical elements

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Software
  • Theoretical Computer Science

Cite this

Kim, D. (2007). A quantitative analysis of memory requirement and generalization performance for robotic tasks. In Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference (pp. 285-292). [1277015] (Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference). https://doi.org/10.1145/1276958.1277015
Kim, DaeEun. / A quantitative analysis of memory requirement and generalization performance for robotic tasks. Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference. 2007. pp. 285-292 (Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference).
@inproceedings{30e1507c51a3467ea498a3a5df6f822e,
title = "A quantitative analysis of memory requirement and generalization performance for robotic tasks",
abstract = "In autonomous agent systems, memory is an important element to handle agent behaviors appropriately. We present the analysis of memory requirements for robotic tasks including wall following and corridor following. The robotic tasks are simulated with sensor modeling and motor actions in noisy environments. In this paper, control structures are based on finite state machines for memory-based controllers, and we use the evolutionary multiobjective optimization approach with two objectives, behavior performance and memory size. For each task, a quantitative approach to estimate internal states with a different number of sensors is applied and the best controllers are evaluated in several test environments to examine their generalization characteristics and efficiency. Finite state machines with a hierarchy of memory are also compared with feedforward neural networks for the behavior performance.",
author = "DaeEun Kim",
year = "2007",
month = "8",
day = "28",
doi = "10.1145/1276958.1277015",
language = "English",
isbn = "1595936971",
series = "Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference",
pages = "285--292",
booktitle = "Proceedings of GECCO 2007",

}

Kim, D 2007, A quantitative analysis of memory requirement and generalization performance for robotic tasks. in Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference., 1277015, Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference, pp. 285-292, 9th Annual Genetic and Evolutionary Computation Conference, GECCO 2007, London, United Kingdom, 07/7/7. https://doi.org/10.1145/1276958.1277015

A quantitative analysis of memory requirement and generalization performance for robotic tasks. / Kim, DaeEun.

Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference. 2007. p. 285-292 1277015 (Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A quantitative analysis of memory requirement and generalization performance for robotic tasks

AU - Kim, DaeEun

PY - 2007/8/28

Y1 - 2007/8/28

N2 - In autonomous agent systems, memory is an important element to handle agent behaviors appropriately. We present the analysis of memory requirements for robotic tasks including wall following and corridor following. The robotic tasks are simulated with sensor modeling and motor actions in noisy environments. In this paper, control structures are based on finite state machines for memory-based controllers, and we use the evolutionary multiobjective optimization approach with two objectives, behavior performance and memory size. For each task, a quantitative approach to estimate internal states with a different number of sensors is applied and the best controllers are evaluated in several test environments to examine their generalization characteristics and efficiency. Finite state machines with a hierarchy of memory are also compared with feedforward neural networks for the behavior performance.

AB - In autonomous agent systems, memory is an important element to handle agent behaviors appropriately. We present the analysis of memory requirements for robotic tasks including wall following and corridor following. The robotic tasks are simulated with sensor modeling and motor actions in noisy environments. In this paper, control structures are based on finite state machines for memory-based controllers, and we use the evolutionary multiobjective optimization approach with two objectives, behavior performance and memory size. For each task, a quantitative approach to estimate internal states with a different number of sensors is applied and the best controllers are evaluated in several test environments to examine their generalization characteristics and efficiency. Finite state machines with a hierarchy of memory are also compared with feedforward neural networks for the behavior performance.

UR - http://www.scopus.com/inward/record.url?scp=34548140150&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548140150&partnerID=8YFLogxK

U2 - 10.1145/1276958.1277015

DO - 10.1145/1276958.1277015

M3 - Conference contribution

AN - SCOPUS:34548140150

SN - 1595936971

SN - 9781595936974

T3 - Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference

SP - 285

EP - 292

BT - Proceedings of GECCO 2007

ER -

Kim D. A quantitative analysis of memory requirement and generalization performance for robotic tasks. In Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference. 2007. p. 285-292. 1277015. (Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference). https://doi.org/10.1145/1276958.1277015