Classes of search:
- Discrete vs. Continuous
- With/without gradient
Discrete search:
- Mapping the search for a plan onto graph search
- States -> Nodes, Actions -> edges, Plans -> paths
- Plan space planning
- BFS/DFS
- Search as queue stack entry
- difference from wind/unwind DFS
- Trees vs graphs (cycle detection)
- Search complexity
- See costs Russel and Novig pp 81
- Costs in a constrained space (e.g. on a plane)
- Search as queue stack entry
- Problem complexity
- Overconstrained, underconstrained, "edge of chaos"
Speeding up search:
- Iterative deepening
- Island
- forward/backward search
- Macros
- Small World Graphs
- Breadth/Depth trade off
- Bias for 'likely' searches
- Path out of local minima
- Pruning
- Abstraction
- Re-representation
- Add a Heuristic