Skip to main content

Project Requirements Checklist

Use this comprehensive checklist to ensure your capstone project meets all requirements before submission.

📋 Core Functionality Checklist​

1. Autonomous Navigation (20 points)​

SLAM Implementation​

  • Map building works in Gazebo/Isaac Sim
  • Localization accuracy within 10cm
  • SLAM algorithm documented (Cartographer, ORB-SLAM3, etc.)
  • Map saved and can be reloaded

Path Planning​

  • Global planner generates paths to goal
  • Local planner avoids obstacles
  • Paths are collision-free
  • Replanning works when obstacles move
  • Successfully navigates 10+ meters
  • Avoids static obstacles (walls, furniture)
  • Avoids dynamic obstacles (moving objects)
  • Reaches goal within 5cm accuracy

Testing Scenarios​

  • Tested in cluttered environment
  • Tested with narrow passages
  • Tested with moving obstacles
  • Success rate >80% over 10 runs

2. Vision-Based Manipulation (25 points)​

Object Detection​

  • Detects at least 5 object classes
  • Detection accuracy >85%
  • Runs in real-time (>10 FPS)
  • Model documented (YOLO, Mask R-CNN, etc.)

Pose Estimation​

  • Estimates 6D pose (position + orientation)
  • Pose estimation error <2cm translation, <5° rotation
  • Works with RGB-D camera data
  • Handles partial occlusions

Grasp Planning​

  • Generates valid grasps for detected objects
  • Collision-free arm trajectories
  • Uses inverse kinematics solver
  • Adjusts grasp based on object geometry

Manipulation Performance​

  • Picks up 3+ different objects
  • Places objects at target locations
  • Grasp success rate >70%
  • No collisions with environment

Testing Scenarios​

  • Pick from table surface
  • Place in container/shelf
  • Handle different object sizes
  • Tested 10+ pick-place cycles

3. Bipedal Locomotion (20 points)​

Walking Controller​

  • Stable walking on flat ground
  • Walking speed adjustable (0.1-1.0 m/s)
  • Smooth gait transitions (stop, start, turn)
  • Uses ZMP or similar stability criterion

Stair Climbing​

  • Climbs at least 3 stairs
  • Stair height 15-20cm (standard)
  • No falls during climb
  • Descends safely

Balance Control​

  • Maintains balance during arm movements
  • Recovers from small pushes (<5N)
  • IMU-based balance feedback
  • No falls during 5-minute operation

Testing Scenarios​

  • Walk 5 meters straight
  • Walk in figure-8 pattern
  • Climb stairs while carrying object
  • Stand on one leg for 3 seconds

4. Conversational Interaction (20 points)​

Speech Recognition​

  • Recognizes voice commands
  • Accuracy >90% in quiet environment
  • Wake word detection ("Hey robot")
  • Uses Whisper or similar ASR

Language Understanding​

  • LLM integration (GPT-4, LLaMA, etc.)
  • Understands 10+ command types
  • Handles natural language variations
  • Context-aware responses

Text-to-Speech​

  • Natural-sounding voice output
  • Speaks acknowledgments
  • Error messages clearly communicated
  • Uses modern TTS (ElevenLabs, Azure, Google)

Command Set (Minimum 10)​

  • "Pick up the [color] [object]"
  • "Move to the [location]"
  • "What do you see?"
  • "Describe your status"
  • "Stop" / "Emergency stop"
  • "Go to the kitchen"
  • "Find the red ball"
  • "Place it on the table"
  • "How many objects do you see?"
  • "What's your battery level?"

Testing Scenarios​

  • Test all commands individually
  • Test command sequences
  • Test with background noise
  • Record command success rate

5. System Integration (15 points)​

Launch System​

  • Single launch file starts all nodes
  • Parameters loaded from YAML files
  • Namespaces used correctly
  • Can launch in simulation or hardware mode

Error Handling​

  • Graceful failures (no crashes)
  • Retry logic for failed operations
  • Logs errors with timestamps
  • Emergency stop works reliably

Performance​

  • All nodes run in real-time
  • CPU usage <80% on target hardware
  • Memory leaks checked (valgrind)
  • Latency <500ms for commands

System Monitoring​

  • ROS 2 diagnostics implemented
  • Battery level monitoring
  • Sensor health checks
  • Status published to dashboard

🌟 Advanced Features (Bonus Points)​

Multi-Modal Perception (+10 points)​

  • Sensor fusion (camera + LiDAR + IMU)
  • Extended Kalman Filter or particle filter
  • Handles sensor failures
  • Improved localization accuracy

Reinforcement Learning (+15 points)​

  • RL policy trained for subtask
  • Training in Isaac Gym or similar
  • Sim-to-real transfer demonstrated
  • Performance comparison with baseline

Human-Robot Collaboration (+10 points)​

  • Hands object to human safely
  • Gesture recognition (MediaPipe, OpenPose)
  • Maintains safe distance from humans
  • Social navigation behaviors

Dynamic Environments (+10 points)​

  • Navigates around moving people
  • Real-time replanning (<1s)
  • Predictive obstacle avoidance
  • Tested with 3+ dynamic obstacles

Real Hardware Deployment (+20 points)​

  • Deployed on physical robot
  • All core features work on hardware
  • Sim-to-real transfer documented
  • Video of hardware demonstration

📦 Deliverables Checklist​

1. Source Code (GitHub Repository)​

Repository Structure​

  • Clear folder organization
  • All ROS 2 packages in src/ directory
  • Launch files in launch/ folders
  • Config files in config/ folders
  • World/model files organized

Code Quality​

  • Follows ROS 2 naming conventions
  • Python code follows PEP 8
  • C++ code follows Google style guide
  • No commented-out code blocks
  • No hardcoded paths

Documentation​

  • README.md with setup instructions
  • ARCHITECTURE.md with system design
  • LICENSE file included
  • Dependencies listed in requirements.txt
  • All packages have package.xml

Version Control​

  • Meaningful commit messages
  • Feature branches (not all on main)
  • .gitignore configured properly
  • No large binary files committed
  • Build artifacts excluded

README.md Required Sections​

  • Project title and description
  • Team members
  • Hardware/software requirements
  • Installation instructions
  • Usage instructions
  • Launch file examples
  • Demo video link
  • Known issues
  • License information

2. Demo Video (3-5 minutes)​

Content Requirements​

  • Introduction with names/title
  • System architecture overview
  • Navigation demonstration
  • Object detection and grasping
  • Locomotion (walking + stairs)
  • Voice command interaction
  • Advanced features (if applicable)
  • Reflections/learnings

Technical Quality​

  • 1080p resolution minimum
  • Clear audio (no background noise)
  • Smooth transitions between clips
  • Text overlays for clarity
  • Picture-in-picture views
  • Uploaded to YouTube or Vimeo
  • Link works and is accessible

Presentation​

  • Professional narration
  • Explains what's happening
  • Highlights key features
  • Shows real-time execution
  • Demonstrates failure recovery
  • Credits used libraries/tools

3. Technical Report (5-10 pages)​

Abstract (0.5 pages)​

  • Problem statement
  • Approach summary
  • Key results
  • Main contributions

Introduction (1 page)​

  • Motivation for project
  • Background on humanoid robotics
  • Related work (2-3 citations)
  • Paper organization

System Architecture (2-3 pages)​

  • Block diagram of system
  • ROS 2 node graph (rqt_graph)
  • Component descriptions
  • Data flow explained
  • Hardware/software stack
  • Design decisions justified

Implementation Details (2-3 pages)​

  • Navigation: Algorithm, parameters
  • Manipulation: Detection model, grasp planner
  • Locomotion: Controller type, gait parameters
  • Conversation: LLM choice, speech pipeline
  • Code snippets for key algorithms
  • Figures/diagrams for clarity

Experimental Results (1-2 pages)​

  • Quantitative metrics table
  • Success rates for each capability
  • Performance graphs (latency, accuracy)
  • Comparison to baseline
  • Failure analysis with examples
  • Statistical significance (if applicable)

Conclusion & Future Work (0.5 pages)​

  • Summary of achievements
  • Challenges overcome
  • Lessons learned
  • Future improvements
  • Broader impact

References​

  • At least 5 citations
  • Proper IEEE/ACM format
  • Mix of papers and documentation
  • All citations used in text

Formatting​

  • IEEE or ACM template used
  • Consistent font and spacing
  • Page numbers included
  • Figures numbered and captioned
  • Tables formatted properly
  • PDF format
  • No spelling/grammar errors

✅ Pre-Submission Checklist​

Testing​

  • Code runs without errors on clean system
  • All dependencies installable
  • Launch files tested
  • Demo recorded successfully
  • Report proofread
  • All links work

Code Cleanup​

  • Removed debug print statements
  • Removed unused imports
  • Removed TODO comments
  • Updated comments to match code
  • Ran code formatter (black, clang-format)
  • Fixed all linter warnings

Repository​

  • README updated
  • Demo video link added
  • All files committed
  • Repo is public or access granted
  • No sensitive data (API keys, passwords)

Documentation​

  • All sections complete
  • Figures have captions
  • References formatted correctly
  • PDF generated successfully
  • File size <10MB

📊 Grading Breakdown​

ComponentPointsYour Score
Navigation20___ / 20
Manipulation25___ / 25
Locomotion20___ / 20
Conversation20___ / 20
Integration15___ / 15
Code Quality10___ / 10
Report10___ / 10
Demo Video10___ / 10
Base Total130___ / 130
Bonus Features+30 max+___
Final Score100%___%

🚀 Tips for Success​

Week 1 (Planning)​

  1. Start early - Don't wait until the last week
  2. Use existing packages - Don't reinvent the wheel
  3. Test incrementally - Get each module working before integration
  4. Ask for help - Use office hours and discussion forums

Week 2 (Implementation)​

  1. Commit often - Version control is your friend
  2. Document as you go - Don't leave it for the end
  3. Record test videos - Useful for debugging and final demo
  4. Handle edge cases - What happens when detection fails?

Week 3 (Documentation)​

  1. Polish the demo - Edit out long pauses
  2. Write clearly - Assume reader is smart but unfamiliar
  3. Include visuals - A picture is worth 1000 words
  4. Proofread everything - Typos hurt professionalism

Common Pitfalls to Avoid​

  • ❌ Hardcoding paths (/home/yourname/...)
  • ❌ Not testing on clean system
  • ❌ Waiting until last minute
  • ❌ Ignoring error messages
  • ❌ Skipping code documentation
  • ❌ Poor video quality
  • ❌ Missing citations in report

📚 Resources​

ROS 2 Best Practices:

Writing Technical Reports:

Video Production:

  • OBS Studio (screen recording)
  • DaVinci Resolve (editing, free)
  • Audacity (audio editing)

Need Help?​

Stuck on a requirement?

  • Review corresponding week's material
  • Check GitHub issues in reference implementations
  • Post on course forum
  • Attend office hours

Technical issues?

  • Search ROS Answers
  • Check package documentation
  • Enable verbose logging for debugging

Ready to build? Head to Autonomous Humanoid Guide 🤖