Learning Outcomes & Assessments
Course Objectives
By the end of this course, you will be able to:
1. Technical Competencies
- ✅ Design and implement ROS 2 systems for robotic control
- ✅ Build digital twins using Gazebo and Unity
- ✅ Deploy perception and navigation pipelines with NVIDIA Isaac
- ✅ Integrate Large Language Models with robotic action systems (VLA)
- ✅ Train reinforcement learning policies for locomotion
- ✅ Perform sim-to-real transfer for physical robot deployment
2. System Integration Skills
- ✅ Architect end-to-end embodied AI systems
- ✅ Debug complex multi-process robotic applications
- ✅ Optimize for real-time performance on edge devices
- ✅ Design hardware-software co-design solutions
3. Research & Innovation
- ✅ Evaluate state-of-the-art Physical AI research papers
- ✅ Identify gaps and propose novel solutions
- ✅ Contribute to open-source robotics projects
Module-Level Learning Outcomes
Module 1: ROS 2 - The Robotic Nervous System
Core Outcomes:
- Implement publisher-subscriber architectures for sensor-actuator loops
- Write Python agents using
rclpyto control robots - Model humanoid robots using URDF/Xacro
- Debug ROS 2 systems using command-line tools (
ros2 topic,ros2 node)
Assessment:
- Lab Reports (3): 40%
- URDF Modeling Assignment: 30%
- Mini-Project (ROS 2 Nervous System): 30%
Module 2: Digital Twin - Simulation & Visualization
Core Outcomes:
- Configure physics simulations (gravity, friction, collision)
- Integrate sensors (LIDAR, cameras, IMU) in Gazebo
- Build Unity-ROS bridges for high-fidelity visualization
- Analyze sim-to-real domain gaps
Assessment:
- Gazebo Lab Series (4): 50%
- Unity Integration Assignment: 20%
- Sim-to-Real Analysis Report: 30%
Module 3: The AI-Robot Brain - NVIDIA Isaac
Core Outcomes:
- Generate synthetic datasets using Isaac Sim
- Implement Visual SLAM and Nav2 for autonomous navigation
- Train RL policies for bipedal locomotion in Isaac Gym
- Deploy models on NVIDIA Jetson hardware
Assessment:
- Isaac Sim Perception Pipeline: 30%
- Navigation Project (Nav2): 30%
- RL Locomotion Training: 40%
Module 4: Vision-Language-Action (VLA) Systems
Core Outcomes:
- Integrate speech recognition (Whisper) with ROS 2
- Use LLMs (GPT-4) for task planning and reasoning
- Fuse multi-modal inputs (vision, speech, sensor data)
- Build conversational robotics interfaces
Assessment:
- Voice Command Lab: 20%
- VLA Agent Assignment: 40%
- Capstone Project: 40%
Capstone Project: Autonomous Humanoid System
Problem Statement
Design a humanoid robot system that:
- Accepts voice commands (e.g., "Clean the table")
- Plans a sequence of actions using an LLM
- Navigates to the target location using Nav2
- Identifies objects using computer vision
- Manipulates objects using inverse kinematics
Evaluation Rubric
| Criterion | Excellent (9-10) | Good (7-8) | Satisfactory (5-6) | Needs Improvement (0-4) |
|---|---|---|---|---|
| System Architecture | Modular, scalable, well-documented | Functional with minor issues | Basic implementation | Incomplete or non-functional |
| Code Quality | Clean, commented, follows ROS 2 best practices | Readable with some inconsistencies | Functional but messy | Poorly structured |
| Demo Performance | Completes all tasks autonomously | Completes most tasks with minor errors | Partial task completion | Fails to demonstrate core functionality |
| Documentation | Comprehensive README, architecture diagrams, video demo | Good documentation, missing some details | Basic documentation | Minimal or missing documentation |
| Innovation | Novel approach or feature | Creative use of existing tools | Standard implementation | No distinguishing features |
Weight: 20% of final grade
Assessment Schedule
| Week | Assessment | Type | Weight |
|---|---|---|---|
| 1 | Environment Setup Report | Lab | 2% |
| 2 | Multi-Node ROS 2 System | Assignment | 5% |
| 3-4 | Voice-to-Motion Controller | Assignment | 5% |
| 5 | Module 1 Mini-Project | Project | 10% |
| 6-7 | Gazebo Sensor Integration | Lab Series | 8% |
| 8 | Sim-to-Real Analysis | Report | 5% |
| 9 | Isaac Sim Perception | Lab | 5% |
| 10 | Nav2 Navigation Project | Project | 10% |
| 11 | RL Locomotion Training | Assignment | 10% |
| 12 | VLA Agent System | Assignment | 10% |
| 13 | Capstone Demo | Final Project | 20% |
| Ongoing | Participation & Peer Review | Continuous | 10% |
Total: 100%
Grading Scale
| Grade | Percentage | Description |
|---|---|---|
| A+ | 95-100% | Outstanding mastery |
| A | 90-94% | Excellent understanding |
| A- | 85-89% | Very good performance |
| B+ | 80-84% | Good performance |
| B | 75-79% | Satisfactory |
| B- | 70-74% | Acceptable |
| C+ | 65-69% | Marginal pass |
| C | 60-64% | Minimum pass |
| F | <60% | Fail |
Academic Integrity
- All code must be original or properly cited
- Open-source libraries are encouraged (with attribution)
- Collaboration is allowed for labs, but assignments and capstone must be individual work
- Plagiarism will result in automatic failure
Support Resources
- Office Hours: TBD (Instructor-led Q&A)
- Discussion Forum: Discord/Slack channel
- Lab Access: Cloud GPU instances or on-campus workstations
- Hardware Loans: Jetson Orin kits available (申請 required)