Name
Abstract | ||
|---|---|---|
Verification of self-adaptive systems is a key area of research within the adaptive systems community. Self-adaptive systems change over time. These changes are frequently based on a stimulus from the outside environment. Adaptive systems may learn how to adapt in real time. The unstable nature of adaptive systems present challenges to testers. This is especially true when portions of an adaptive system achieve a stable testable state, and other portions do not. This study researches how to identify stable states within a Bayesian self-adaptive system. This paper presents an example self-adapting robot system whose function is avoiding obstacles within a simulation environment. It contains a static non-avoiding simulated robot, a rule-based obstacle avoiding robot, and a self-adapting simulated robot. The paper describes the performance different simulated robots and compares them against one another. This research analyzes the emerging set of Bayesian posterior probabilities in order to discover the point in time where the self-adaptive robot system achieves homeostasis. Engineers may use this point to execute verification or validation processes with minimal fear that new adaptation will interfere with results. |
Year | DOI | Venue |
|---|---|---|
2014 | 10.1016/j.procs.2014.09.047 | Procedia Computer Science |
Keywords | Field | DocType |
Bayesian,verification,homeostasis,trigger,event | Obstacle,Data mining,Robotic systems,Computer science,Stable states,Adaptive system,Posterior probability,Self adaptive,Artificial intelligence,Robot,Machine learning,Bayesian probability | Conference |
Volume | ISSN | Citations |
36 | 1877-0509 | 1 |
PageRank | References | Authors |
0.34 | 2 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Brian J. Phillips | 1 | 2 | 1.38 |
| Mark Blackburn | 2 | 45 | 6.57 |