jfowkes
Member
- Joined
- 20 Jul 2017
- Messages
- 942
A comment in the EWR construction updates thread got me thinking about route learning:
Basically, my questions are how much route learning is done in simulators now, and how much could be done in future?
I know nothing about the topic, so I don't know if simulators are too limited to really give you the real world fidelity you need for good route learning.
I imagine you'd need a number of real-world runs anyway. But if a simulator can get you 90% of the way there, plus getting you learning in different weather conditions etc., I imagine this could substantially reduce the amount of actual runs required? Seems especially useful for new lines like EWR, where the route doesn't actually exist yet.
Except that a more logical timescale would have been to have been in a position to start this process so that it finishes when the assets become available for use. (Accept that route learning would need to wait until then, but I would have thought that would be a relatively small part of the process.) Unfortunately I don't think the DfT do "logical" very often, and in particular the delays in choosing a TOC to run it won't have helped.
Basically, my questions are how much route learning is done in simulators now, and how much could be done in future?
I know nothing about the topic, so I don't know if simulators are too limited to really give you the real world fidelity you need for good route learning.
I imagine you'd need a number of real-world runs anyway. But if a simulator can get you 90% of the way there, plus getting you learning in different weather conditions etc., I imagine this could substantially reduce the amount of actual runs required? Seems especially useful for new lines like EWR, where the route doesn't actually exist yet.