Task
As a team, merge your six lab results into one controller + estimator stack (no per-lab
hacks left inside), and fly a complete mission in
quadsim:
take off → visit ≥ 3 waypoints → land, closing the loop on the estimated state,
with a wind disturbance enabled. Then — the analytical heart of the assignment — identify one
concrete limitation your classical autopilot shows, demonstrate it with a plot or metric, and
argue which intelligent-control upgrade (MPC or RL, Weeks 11–14) should fix it. That limitation
becomes your team's final-project problem statement.
What to submit (one zip per team)
- Code — the integrated stack (
student.py+ estimator wiring + mission scriptmission.py) that reproduces your demo with one command - Report (PDF, 5–8 pages) — architecture diagram, mission result (3-D flight path + tracking/estimation plots), the limitation demonstration, and the proposed upgrade with justification
- Slides — the ~10-minute in-class presentation (Week 10)
- Contribution statement — half a page: who did what, signed by every member
Pass bar — expected results
- Mission completes from estimated state under wind: take-off, ≥ 3 waypoints reached (within 0.3 m each), landing — no crash, no ground-truth leakage into the controller
- The limitation is shown, not asserted — a figure or number that a Weeks 11–14 technique can plausibly improve
- One-command reproduction:
PYTHONPATH=. python mission.py
Assessment criteria
| Criterion | Weight | What earns full marks |
|---|---|---|
| Mission success (estimated state + wind) | 30% | The integrated autopilot completes the full mission as specified, reproduced by re-running the team's code. |
| Integration quality | 20% | One coherent stack — consistent conventions, the estimator feeding the controller, no duplicated or dead per-lab code. |
| Limitation analysis → upgrade proposal | 20% | The limitation is demonstrated with evidence, correctly diagnosed, and the proposed MPC/RL upgrade logically targets it. |
| Presentation (Week 10, ~10 min) | 15% | Clear story in the time limit: what was built → what it does → where it breaks → what comes next; every member speaks. |
| Report | 10% | 5–8 pages, labelled figures, honest reporting of what does and does not work. |
| Contribution statement | 5% | Specific and credible; grades may be individually adjusted for grossly unequal contributions. |
Rules
- Team work — this assignment and the final project run in the same teams (register your team by Week 5).
- Cross-team idea exchange is fine; code and text stay within the team. AI-assistant policy as in the lab assignments: declare use, be able to explain every line.
- Late submission: −20% per day, not accepted after 3 days (presentation slot cannot be made up — absence without cause scores 0 on the presentation criterion).
📤 Submit via the MUST LMS
What
one zip per team (code + report + slides + contribution statement)
Filename
MLTE03_CaseStudy_Team<K>.zipWhere
MUST LMS → MLTE03 → Team Case Study dropbox LMS link — TO FILL
Deadline
date / time — TO FILL (presentation in the Week 10 session)