What is covered, week by week, and which topics each lecture draws on.
Lectures#
Twelve Monday lectures, one question each. The first three build a search you can trust; week 4 makes it respect a battery; weeks 5 and 6 turn one route into a schedule; weeks 7 and 8 divide the work between drones that share two charging pads; weeks 9 to 12 make five plans executable in one airspace. Week 9 carries the deck. Each page contains learning goals, a model boundary, a worked derivation or trace, a figure, a checkpoint with an explanation, and a link to the matching tutorial. Read the lesson, make a prediction, then use the tutorial to test it.
Model legal edges, trace Dijkstra, and explain A*.
Week 1: One order, a complete task
Why a line from the kitchen to the customer is not yet a problem definition: graph, state, legal edges, goal and cost.
Week 2: Dijkstra: when to stop
Tentative distances, settled nodes and relaxation; why the goal being discovered is not the goal being found.
Week 3: A*: a useful estimate
f = g + h, lower bounds with units, admissible versus consistent, and what re-expansion is for.
Keep time and energy labels for a complete delivery.
Week 4: A route that can return
Payload, energy and the return leg; time–energy labels, dominance, and why the fastest arrival is not enough.
Declare an objective and test local improvement.
Week 5: Which dinner goes first?
The timetable recurrence, and four objectives that are not the same objective.
Week 6: When swaps stop helping
Plans as permutations, a swap neighbourhood, strict improvement, and what a local optimum does and does not mean.
Assign capable drones and share the charging pads.
Week 7: Choose a capable drone
One sequence per drone, a feasibility matrix, earliest-completion greedy assignment and cross-drone migration.
Week 8: Who gets the charging pad?
Two charging pads as a shared resource: discrete events, capacity intervals, the queue recurrence and how a change reaches a third drone.
Reserve shared airspace and feed delays back into the plan.
Week 9: Two drones, one corridor
Defining a conflict, putting time into the state, half-open occupancy intervals, and why visited[node] deletes waiting.
Week 10: Search around a reservation
A* over (place, phase, time) under committed reservations; prioritised planning and what it does and does not guarantee.
Week 11: Let delays change the assignment
Assignment, routes, resources and the objective as one evaluation; full re-evaluation of every candidate; what a decomposed method can claim.
Week 12: Audit the twenty dinners
From spatial nodes to resource labels to space-time states; proven properties, tested results and untested generalisations.