Chapter 10 algorithm implementations

[diegoalejogm]

I was wondering how you expect to have implemented the PPDL definitions of the problems in chapter 10. I've implemented them in my AI nanodegree approach using aima.util expressions, but I have no clarity on what do you expect it to do in this source.

Should it be able to be run? If so, what kind of search algorithms (DFS,BFS,etc) should be used.

What I mean is that it isn't as explicit as coding an algorithm, because those definitions aren't.

[norvig]

@bcorfman has a good example of how to do this: https://github.com/bcorfman/aima-python/blob/master/planning.py

[diegoalejogm]

Thanks @norvig.

I tried running implementations found in aimacode's repo and those won't work. Conversely, just tried @bfcorfman 's code and those ran correctly, outputting the expected values. So if they're is solved already, why hasn't there been a pull request?

I'd be happy to re-implement it if necessary. Actually I've got a similar approach here: https://github.com/diegoalejogm/AI-Nanodegree/blob/master/planning-search/my_air_cargo_problems.py

But again, if everything is done, why isn't it already in the repo?

[dmeoli]

The Chapter 10 (3e) classical-planning algorithms are implemented and tested in planning.py: PDDL-style PlanningProblem/Action with the canonical domains (air_cargo, spare_tire, three_block_tower, have_cake_and_eat_cake_too, shopping_problem, simple_blocks_world), forward and backward state-space search (ForwardPlan, BackwardPlan, runnable with astar_search), and the planning graph + GraphPlan (with mutex handling, recently reworked in #1324). Covered by the air_cargo/spare_tire/three_block_tower/graph_plan/forward_plan/backward_plan tests. The original 'won't work' complaint no longer applies. Closing as implemented.