Assessments & Capstone Project

Learning Objectives

By the end of this section, you will understand:

• The weekly assessment structure and how it contributes to your grade
• Lab assignment requirements for each module
• The complete capstone project specification and grading rubric
• Submission deadlines and evaluation criteria

Weekly Assessment Structure

Each week includes a short quiz (10 questions, 15 minutes) covering:

• Core concepts from the week's lectures
• Code comprehension questions (read a ROS 2 node, identify what it does)
• Short answer: "Explain the difference between a ROS 2 Topic and a Service"

Quiz Grading

Component	Weight
Weekly Quizzes (13 × 5 pts)	65 pts
Lab Assignments (4 modules)	100 pts
Final Capstone Project	150 pts
Participation	35 pts
Total	350 pts

Lab Assignments

Lab 1: ROS 2 Publisher-Subscriber (Module 1)

Objective: Build a complete ROS 2 workspace with:

• A sensor data publisher (joint states at 50 Hz)
• A data processor subscriber (moving average filter)
• A custom URDF of a 2-DOF arm

# Expected workspace structure
my_humanoid_ws/
├── src/
│   └── my_robot_pkg/
│       ├── my_robot_pkg/
│       │   ├── joint_publisher.py
│       │   └── joint_subscriber.py
│       ├── urdf/
│       │   └── two_dof_arm.urdf
│       └── launch/
│           └── robot.launch.py
├── install/
└── build/

Grading:

• Publisher runs without errors: 10 pts
• Subscriber logs averaged data: 10 pts
• URDF loads in RViz2: 5 pts

Lab 2: Gazebo Simulation (Module 2)

Objective: Spawn your URDF in Gazebo Harmonic, attach a simulated camera, and visualize the camera feed in RViz2.

Grading: Simulation runs + camera stream visible: 25 pts

Lab 3: Nav2 Navigation (Module 3)

Objective: Use Isaac ROS VSLAM to map a simulated room (Isaac Sim), then navigate to 3 waypoints autonomously using Nav2.

Grading:

• SLAM map generated: 10 pts
• All 3 waypoints reached: 15 pts

Lab 4: VLA Pipeline Prototype (Module 4)

Objective: Build a minimal VLA pipeline that accepts a typed text command (no audio hardware required) and generates a ROS 2 action sequence using GPT-4o-mini.

Grading: Command → Plan → ROS 2 action execution: 25 pts

Final Capstone Project

"Autonomous Humanoid Assistant"

Build a system where:

1. A user speaks: "Go to the table and pick up the red cup"
2. Whisper transcribes the command
3. GPT-4 generates an action plan
4. The robot (simulation or physical) navigates to the table
5. Computer vision detects and localizes the cup
6. The robot arm grasps the cup
7. The robot returns to home position and announces completion

Capstone Grading Rubric (150 pts)

Component	Max Points	Passing Threshold
Voice recognition & transcription	20	14
GPT task planning (valid JSON output)	25	17
Robot navigation to target	30	20
Object detection accuracy	25	17
Successful grasp	25	17
Demo video (90 seconds, clear narration)	15	10
Code quality & documentation	10	7

Submission Requirements

1. GitHub repository — clean, with README and live deployment link
2. 90-second demo video — narrated screen/robot recording
3. Written report (500 words) — architecture decisions, challenges, lessons learned
4. Live demo (optional) — synchronous presentation for bonus 10 pts

Deadlines

• Lab 1: End of Week 5
• Lab 2: End of Week 7
• Lab 3: End of Week 10
• Lab 4: End of Week 12
• Capstone: End of Week 13 (Final Day)

---

Congratulations on completing the Physical AI & Humanoid Robotics curriculum!

You now have the foundations to build, program, and deploy autonomous humanoid robot systems using the latest AI tools and robotics frameworks.