Abstract
An important capability of realtime search is learning, that is, as in LRTA*, the solution path converges to an optimal path by repeating problem solving trials1. In this chapter, we will focus not on the performance of the first problem solving trial, but on the learning process to converge to an optimal solution. Previous research on realtime search mainly addressed problem solving performance, and did not pay much attention to the learning process. This chapter is the first to point out that the following problems are incurred when repeatedly applying LRTA* to solve a problem.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Rights and permissions
Copyright information
© 1997 Kluwer Academic Publishers
About this chapter
Cite this chapter
Shimbo, M. (1997). Controlling Learning Processes. In: Real-Time Search for Learning Autonomous Agents. The Springer International Series in Engineering and Computer Science, vol 406. Springer, Boston, MA. https://doi.org/10.1007/978-0-585-34507-9_2
Download citation
DOI: https://doi.org/10.1007/978-0-585-34507-9_2
Publisher Name: Springer, Boston, MA
Print ISBN: 978-0-7923-9944-5
Online ISBN: 978-0-585-34507-9
eBook Packages: Springer Book Archive