Skip to main content

Assessments

Overview​

This course employs a project-based assessment approach designed to reinforce theoretical concepts with practical application. Each assessment builds upon the skills and knowledge acquired in previous modules.

Assessment Breakdown​

AssessmentModuleWeightDue Week
ROS 2 Package ProjectModule 120%Week 4
Digital Twin SimulationModule 225%Week 7
Isaac Perception PipelineModule 325%Week 12
Capstone Humanoid ProjectModule 430%Week 13

1. ROS 2 Package Project (20%)​

Objective​

Demonstrate proficiency in fundamental ROS 2 concepts by designing, implementing, and testing a modular ROS 2 package.

Requirements​

Minimum Components:

  • At least 3 nodes with clear responsibilities
  • Topic-based communication (minimum 2 topics)
  • At least 1 service OR 1 action
  • Proper package structure with dependencies

Technical Requirements:

  • Python or C++ implementation
  • Launch file to start all nodes
  • Configuration files (YAML) for parameters
  • Error handling and logging

Documentation:

  • README with system overview
  • Architecture diagram showing node communication
  • Code comments explaining functionality
  • Usage instructions

Example Project Ideas​

  1. Simulated Robot Controller

    • Keyboard input node
    • Velocity command publisher
    • Simulated robot state node
    • Visualization node

  2. Sensor Data Pipeline

    • Sensor data generator
    • Data processing node
    • Data visualization node
    • Data logging service

  3. Multi-Agent Coordinator

    • Multiple robot simulation nodes
    • Coordination logic node
    • Collision avoidance service
    • Status monitoring node

Evaluation Criteria​

CriterionWeightDescription
Architecture30%Clear node design, proper separation of concerns
Communication30%Correct use of topics, services, or actions
Code Quality20%Clean code, documentation, error handling
Functionality20%System works as intended, meets requirements

Submission​

  • GitHub repository link
  • README with setup instructions
  • 2-3 minute demonstration video
  • Brief report (2-3 pages) explaining design decisions

2. Digital Twin Simulation (25%)​

Objective​

Develop and interact with a functional digital twin of a humanoid robot in a simulated environment.

Requirements​

Robot Model:

  • URDF/XACRO description with minimum 6 DOF
  • Accurate mass and inertia properties
  • Visual and collision geometries
  • At least 2 sensor types (e.g., camera + IMU)

Simulation Environment:

  • Gazebo OR Unity world
  • Minimum 3 environmental obstacles
  • Textured ground plane
  • Proper lighting

Control Implementation:

  • Joint position or velocity control
  • Keyboard or GUI control interface
  • Stable operation for at least 60 seconds
  • Emergency stop functionality

Sensor Integration:

  • ROS 2 topics publishing sensor data
  • Sensor visualization (RViz or custom)
  • Data logging capability

Deliverables​

  1. Code Package

    • Complete ROS 2 package
    • URDF/XACRO files
    • World/scene files
    • Launch files

  2. Documentation

    • System architecture diagram
    • Robot specifications (DOF, sensors, mass)
    • Setup and launch instructions
    • Known limitations

  3. Demonstration

    • 3-5 minute video showing:
      • Robot model in simulation
      • Control interface
      • Sensor data visualization
      • Basic movements

Evaluation Criteria​

CriterionPointsDescription
Model Quality25Accurate kinematics, proper URDF structure
Simulation Fidelity25Realistic physics, sensor behavior
Control Implementation25Responsive, stable control
Documentation15Clear, complete, professional
Demonstration10Effective showcase of capabilities

3. Isaac Perception Pipeline (25%)​

Objective​

Implement and evaluate a perception pipeline using NVIDIA Isaac Sim for a humanoid robot.

Requirements​

Part 1: VSLAM Implementation (40%)

  • Isaac Sim scene with humanoid robot
  • Camera sensor configuration (RGB-D or stereo)
  • VSLAM algorithm (ORB-SLAM3, RTAB-Map, or equivalent)
  • Mapping of unknown environment
  • Real-time localization

Part 2: Nav2 Integration (40%)

  • Nav2 configuration for bipedal locomotion
  • Costmap setup with appropriate parameters
  • Global and local planners configured
  • Goal-oriented navigation demonstration
  • Dynamic obstacle handling

Part 3: Performance Analysis (20%)

  • Map quality metrics (coverage, accuracy)
  • Localization error analysis
  • Navigation success rate
  • Computational performance metrics

Deliverables​

  1. Implementation

    • Isaac Sim scene files
    • ROS 2 packages for VSLAM and Nav2
    • Configuration files
    • Launch files

  2. Analysis Report (5-7 pages)

    • Methodology description
    • Experimental setup
    • Results with graphs and tables
    • Discussion of findings
    • Challenges and solutions

  3. Demonstration Video (5-7 minutes)

    • System overview
    • VSLAM mapping process
    • Navigation to multiple goals
    • Obstacle avoidance
    • Performance metrics visualization

Evaluation Criteria​

CriterionWeightDescription
VSLAM Accuracy25%Map quality, localization precision
Nav2 Configuration25%Appropriate parameters for humanoid
Navigation Performance25%Success rate, path quality, obstacle handling
Analysis & Documentation25%Thorough evaluation, clear reporting

4. Capstone Humanoid Project (30%)​

Objective​

Integrate knowledge and skills from all modules to develop an intelligent humanoid robot capable of performing a complex, multi-step task in a simulated environment.

Requirements​

System Integration:

  • ROS 2 communication framework
  • Isaac Sim simulation environment
  • VSLAM for localization
  • Nav2 for navigation
  • VLA pipeline (Vision-Language-Action)

Core Capabilities:

  1. Locomotion (25%)

    • Stable bipedal walking
    • Navigate to specified locations
    • Avoid dynamic obstacles
    • Recovery from disturbances

  2. Perception (20%)

    • Visual SLAM
    • Object detection and recognition
    • Scene understanding
    • Spatial reasoning

  3. Voice Interface (20%)

    • Speech recognition (Whisper)
    • Natural language command parsing
    • Spoken feedback (text-to-speech)
    • Dialogue management

  4. Task Planning (20%)

    • GPT-based task decomposition
    • Multi-step plan execution
    • Error handling and recovery
    • Adaptive replanning

  5. Manipulation (15%)

    • Pick and place objects
    • Grasp planning
    • Collision avoidance
    • Force control (if applicable)

Task Scenarios (Choose One)​

  1. "Clean up the living room"

    • Detect misplaced objects
    • Plan pickup sequence
    • Navigate and grasp objects
    • Place in designated locations

  2. "Bring me a water bottle from the kitchen"

    • Navigate to kitchen
    • Locate and identify water bottle
    • Grasp and carry safely
    • Return and hand over to person

  3. "Set the table for dinner"

    • Understand table setting requirements
    • Locate plates, utensils, glasses
    • Arrange in proper positions
    • Handle multiple objects

  4. Custom Task (requires approval)

    • Propose your own complex task
    • Must integrate all required capabilities
    • Submit proposal by Week 12

Deliverables​

  1. System Implementation

    • Complete ROS 2 workspace
    • Isaac Sim scene and robot model
    • VLA pipeline code
    • Configuration files

  2. Documentation (10-15 pages)

    • Executive summary
    • System architecture
    • Component descriptions
    • Integration approach
    • Challenges and solutions
    • Future improvements

  3. Demonstration Video (5-7 minutes)

    • System overview
    • Task execution (full run)
    • Key features highlighted
    • Challenges encountered
    • Results and discussion

  4. Final Presentation (15 minutes + 5 min Q&A)

    • Problem statement
    • Technical approach
    • System architecture
    • Live demonstration OR video
    • Results and evaluation
    • Lessons learned

Evaluation Criteria​

CriterionWeightDescription
System Integration25%All modules work together seamlessly
Task Completion25%Successfully completes assigned task
Robustness20%Handles errors, edge cases, disturbances
Innovation15%Creative solutions, extensions beyond requirements
Documentation & Presentation15%Clear, comprehensive, professional

Submission Guidelines​

General Requirements​

  • Code: Well-commented, follows style guidelines
  • Documentation: PDF format, proper citations (APA 7th edition)
  • Videos: MP4 format, 1080p minimum, clear audio
  • Repository: Public GitHub repository with README

Late Submission Policy​

  • Within 24 hours: 10% penalty
  • 24-48 hours: 25% penalty
  • Beyond 48 hours: Not accepted (except with prior approval)

Academic Integrity​

  • All work must be your own
  • Properly cite all sources and references
  • Collaboration is encouraged for learning, but submissions must be individual
  • Use of AI tools (ChatGPT, Copilot) must be disclosed

Grading Scale​

GradePercentageDescription
A90-100%Exceptional work, exceeds expectations
B80-89%Good work, meets all requirements
C70-79%Satisfactory work, meets most requirements
D60-69%Minimal work, meets some requirements
F<60%Unsatisfactory work

Navigation:
← Weekly Schedule | Hardware Architecture →