Planning the course¶

A shape that works¶

The course is twelve modules in a fixed order, and it gets sharply harder at Module 6. A comfortable plan over a term is one module per 90-minute session, with two sessions for Modules 1, 6 and 9.

Session	Module	Prepare
1	0 — Introduction	Accounts. Expect sign-in to take ten minutes.
2–3	1 — Assembly	Kits or printed parts, screwdrivers, hex keys, space to lay parts out.
4	2 — Connection and anatomy	Arms built; USB tested on the lab machines.
5	3 — Calibration	Nothing extra. Budget the whole session; it is the foundation.
6	4 — Teleoperation	A cylinder to move; clear desks.
7	5 — Pick and place	An object to pick.
8–9	6 — End-effector work	A pencil per arm, A4 paper. The longest module by far.
10	7 — Computer vision	Webcams, optional. A blue cube.
11	8 — Tracking and PID	A cube and a cylinder that look similar in colour.
12–13	9 — Neural-network recognition	Cube, cylinder and a nut; good even lighting if using real cameras.
14	10 — ArUco markers	Nothing extra.
—	11 — Final project	In development; plan your own capstone.

If your term is shorter, the natural cut is after Module 6: it is also where the certificate threshold sits — the first seven modules earn it.

What each module really asks of a learner¶

0 — Introduction. Reading and five safety checkpoints. The only module with multiple-choice questions. Do not skip it; it is where the safety rules are established, and you will want to refer back to them.

1 — Assembly. Mechanical work, then the part that catches everyone: assigning motor IDs one motor at a time. Plan for it. A pair that attaches all six motors and then tries to assign IDs will be stuck until an adult explains why.

2 — Connection and anatomy. Gentle. Vocabulary and the 3D view. Good session to let pairs work unsupervised.

3 — Calibration. Conceptually the most important session of the course. A sloppy HOME pose here produces wrong angles for the next ten modules, and learners cannot diagnose that themselves. Walk the room and check poses before anyone presses Calibrate.

4 — Teleoperation. First real driving. The workspace exercise — probing which points are reachable — is the one they remember.

5 — Pick and place. Grip force, an emergency-stop drill, then recording a sequence and replaying it autonomously. The first time the arm does something on its own; leave time to enjoy it.

6 — End-effector work. Two calibrations, three motion types, and programming a signature letter by letter. Expect two sessions. The four-point tool calibration is fiddly and worth demonstrating once at the front.

7 — Computer vision. No arm needed. Runs against a virtual scene or a real webcam. A good session to schedule when hardware is short.

8 — Tracking and PID. The one with real control theory. Tuning three gains by hand teaches more in twenty minutes than a lecture does; let them make it oscillate.

9 — Neural-network recognition. Sixty images collected and labelled by hand. It is long and it is deliberately tedious in the right way — the point being that data work is most of machine learning. The lesson rejects a lazy virtual dataset shot from one pose.

10 — ArUco markers. Short and satisfying. Builds a marker, decodes it, then aims at one by ID.

Pacing without a progress dashboard¶

You cannot see the group's progress from your own screen. Two habits make that a non-issue:

Start each session with a show-of-dashboards. Everyone opens their dashboard; you read the room in thirty seconds.
Anchor on module boundaries, not steps. "Everyone finishes Module 4 this week" is checkable at a glance; "everyone reaches step 5" is not.

Where to put your own material¶

The platform cannot hold homework, assignments or your own lessons. What works well alongside it:

A short written reflection per module — what broke, what fixed it.
The artefacts: an exported path, a photograph of the arm's signature on paper, a screenshot of a tuned PID response, the labelled dataset.
A capstone of your own design in place of the unbuilt Module 11.