Skip to content

Latest commit

 

History

History
416 lines (371 loc) · 21.1 KB

File metadata and controls

416 lines (371 loc) · 21.1 KB

3D Geometry & Visualization Cheatsheet for Robotics

1. Coordinate Systems & Transformations

Coordinate Systems

    Z                 Z
    |                 |
    |                 |    
    |____X       Y____|
   /               /
  /               /
 Y               X
ROS Standard      Camera Standard
  • Right-hand rule: Curl fingers from x to y, thumb points to z
  • Common conventions:
    • ROS standard: x forward, y left, z up
    • Camera standard: z forward, x right, y down
    • OpenGL standard: x right, y up, z backward

Transformation Representation

Homogeneous Transformation Matrix

┌                  ┐
│ r11  r12  r13  tx │
│ r21  r22  r23  ty │
│ r31  r32  r33  tz │
│  0    0    0   1  │
└                  ┘
  Rotation    Translation

Euler Angles Visualization

      Z                 Z'                  Z''
      |                 |                   |
      |                 |                   |
Roll  |     Pitch       |      Yaw          |
→     |     →           |      →            |
      O----> X         O'----> X'          O''----> X''
     /                 /                   /
    /                 /                   /
   Y                 Y'                  Y''

Quaternion Parameters

q = w + xi + yj + zk

Unit quaternion: w² + x² + y² + z² = 1

2. Point Cloud Visualization & Processing

Point Cloud Visualization Types

Raw Points           Colored by Height      Surface Normals
   .  .                  .  .                  .  .
 .      .  ↑           .      .  ↑           .  ↑   .  ↑
.    .    .  |        .    .    .  |        . /    . /
 .     .     | Z       .     .     | Z       /  .    /
   .  .      |           .  .      |        ↑     ↑

Point Cloud Organization

KD-Tree                  Voxel Grid               Octree
    │                   ┌─┬─┬─┬─┐              ┌───┬───┐
    ▼                   ├─┼─┼─┼─┤              │   │   │
   / \                  ├─┼─┼─┼─┤              ├───┼───┤
  /   \                 ├─┼─┼─┼─┤              │   │   │
 /     \                └─┴─┴─┴─┘              └───┴───┘

Common Visualization Color Schemes

Type Use Case Example
Height map Terrain visualization Blue (low) → Green → Yellow → Red (high)
Intensity LiDAR reflectivity Grayscale (0-255)
RGB Camera-colored points Natural color
Segmentation Object classification Unique color per class
Normal Surface orientation RGB = XYZ components of normal

Segmentation Visualization

Raw Point Cloud       Segmented Objects        Instance Masks
   .  .  .  .            .  .  .  .             .  .  .  .
 .  .  .  .  .          .  .  .  .  .           .  .  .  .  .
.  .  .  .  .  .     →  1  1  2  2  3  3    →   🟥 🟥 🟦 🟦 🟨 🟨
 .  .  .  .  .          .  .  .  .  .           .  .  .  .  .
   .  .  .  .            .  .  .  .             .  .  .  .

3. SLAM Visualization Techniques

SLAM Components Visualization

              ┌───────────────┐
              │ Input Sensors │
              └───────┬───────┘
                      │
        ┌─────────────┴─────────────┐
        ▼                           ▼
┌───────────────┐           ┌───────────────┐
│Feature Extract│◄──────────►│Pose Estimation│
└───────┬───────┘           └───────┬───────┘
        │                           │
        ▼                           ▼
┌───────────────┐           ┌───────────────┐
│   Mapping     │◄──────────►│ Loop Closure  │
└───────────────┘           └───────────────┘

Feature Visualization

Raw Image                 Features Detected          Feature Matching
┌──────────────┐         ┌──────────────┐         ┌─────────┬─────────┐
│              │         │   •    •     │         │ •    •  │  •    • │
│              │    →    │       •   •  │    →    │    •    │     •   │
│              │         │  •  •        │         │  •   •  │   •   • │
└──────────────┘         └──────────────┘         └─────────┴─────────┘
                            FAST corners           Correspondences (lines)

Trajectory & Map Visualization

                   Estimated Path
                   .
                  / \
                 /   \
                /     \
Ground Truth   /       \    Loop
─────────────>•         \   Closure
               \         \    |
                \         \   ↓
                 \         \ /
                  \_________•

Factor Graph Visualization

    ┌───┐     ┌───┐     ┌───┐     ┌───┐
    │ X₁│─────│ X₂│─────│ X₃│─────│ X₄│  ← Robot poses
    └─┬─┘     └─┬─┘     └─┬─┘     └─┬─┘
      │         │         │         │
      │         │         │         │     ← Observations
      ▼         ▼         ▼         ▼
    ┌───┐     ┌───┐     ┌───┐     ┌───┐
    │ L₁│     │ L₂│     │ L₃│     │ L₄│  ← Landmarks
    └───┘     └───┘     └───┘     └───┘
      ▲                             ▲
      │                             │
      └─────────────────────────────┘     ← Loop closure

4. Sensor Fusion Visualization

Multi-Sensor Fusion Diagram

┌──────────┐    ┌──────────┐    ┌──────────┐
│  Camera  │    │   LiDAR  │    │   IMU    │
└────┬─────┘    └────┬─────┘    └────┬─────┘
     │               │               │
     ▼               ▼               ▼
┌────┴─────┐    ┌────┴─────┐    ┌────┴─────┐
│  Feature │    │   Point  │    │  Motion  │
│Extraction│    │  Cloud   │    │Estimation│
└────┬─────┘    └────┬─────┘    └────┬─────┘
     │               │               │
     └───────────────┼───────────────┘
                     ▼
              ┌──────────────┐
              │ Fusion Filter│
              │   (EKF/UKF)  │
              └──────┬───────┘
                     ▼
              ┌──────────────┐
              │  Fused State │
              │  Estimation  │
              └──────────────┘

Kalman Filter Visualization

          Prior                 Measurement              Posterior
       Distribution             Distribution             Distribution
          /│\                      /│\                      /│\
         / │ \                    / │ \                    / │ \
        /  │  \                  /  │  \                  /  │  \
       /   │   \                /   │   \                /   │   \
      /    │    \              /    │    \              /    │    \
     /     │     \            /     │     \            /     │     \
    /      │      \          /      │      \          /      │      \
───┴───────┴───────┴────  ───┴───────┴───────┴────  ───┴───────┴───────┴────
          x₁                       x₂                   x₃ (Combined)

Particle Filter Visualization

1. Initial Particles     2. Motion Update      3. Measurement Update
    • • • • • • •           ↗ ↗ ↗               • • • •
    • • • • • • •          ↗ ↗ ↗ ↗            • • • • • •
    • • • ⊙ • • •    →    ↗ ↗ ⊙ ↗ ↗      →   • • • ⊙ • •
    • • • • • • •          ↗ ↗ ↗ ↗               • • •
    • • • • • • •           ↗ ↗ ↗                  •

⊙ = True position        Arrows = Motion      Size = Weight

5. Camera-to-World Transformation Pipeline

Camera Model Visualization

       3D World               Image Plane
          P(X,Y,Z)               p(u,v)
             •                     •
             │                     │
             │                     │
             │      Optical       │
             │       Axis         │
             │         │          │
             ▼         ▼          ▼
          Z  │         │          │
           ╲ │         │          │
            ╲│         │          │
      Camera ⊙─────────┼──────────┼─────► X
      Center           │          │
                  focal length    │
                                  │
                                  ▼
                                  Y

Projection Matrix Visualization

┌   ┐   ┌                 ┐ ┌   ┐
│ u │   │ fx   0   cx   0 │ │ X │
│ v │ = │ 0   fy   cy   0 │ │ Y │
│ 1 │   │ 0    0    1   0 │ │ Z │
└   ┘   └                 ┘ │ 1 │
                            └   ┘

Intrinsics K            3D Point

Transformation Chain Visualization

   Object           Camera           Robot           World
Coordinate        Coordinate       Coordinate      Coordinate
   System           System           System          System
     ↓                ↓                ↓               ↓
  [X_obj]    →     [X_cam]    →     [X_rob]    →    [X_world]
     ↑                ↑                ↑               ↑
     └────────┐      └────────┐       └───────┐       │
              │               │                │       │
         T_obj_cam        T_cam_rob       T_rob_world

6. 3D Perception Debugging Visualization

Pipeline Component Visualization

┌──────────────┐  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐
│  Raw Sensor  │→ │  Preprocess  │→ │   Feature    │→ │   Object     │
│    Data      │  │  & Filter    │  │  Extraction  │  │  Detection   │
└──────┬───────┘  └──────┬───────┘  └──────┬───────┘  └──────┬───────┘
       │                 │                 │                 │
       ▼                 ▼                 ▼                 ▼
┌──────────────┐  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐
│   Raw PCL    │  │  Filtered    │  │  Extracted   │  │  Bounding    │
│  Visualizer  │  │    PCL       │  │  Features    │  │   Boxes      │
└──────────────┘  └──────────────┘  └──────────────┘  └──────────────┘

Error Visualization Methods

Ground Truth         Estimated           Error Heatmap
┌──────────────┐   ┌──────────────┐   ┌──────────────┐
│              │   │              │   │▓▓▒▒▒▒        │
│     ┌───┐    │   │    ┌───┐     │   │▓▓▒▒▒▒        │
│     │   │    │   │    │   │     │   │▓▓▒▒          │
│     └───┘    │ - │    └───┘     │ = │▒▒            │
│              │   │              │   │              │
└──────────────┘   └──────────────┘   └──────────────┘

Common Debug Visualizations

Visualization Type Purpose Display Method
Confidence heatmap Show detection certainty Color gradient overlay
Wireframe overlay Show 3D model alignment Edges on image
Confusion matrix Evaluate classification Color-coded grid
Occlusion visualization Show hidden areas Transparency/hatching
Trajectory comparison Evaluate localization Multiple path lines

7. System Integration & Visualization Architecture

Data Flow Visualization

┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐
│  Sensor  │───►│ Perception│───►│ Planning │───►│ Control  │
│  Fusion  │    │  Stack   │    │  Stack   │    │  Stack   │
└────┬─────┘    └────┬─────┘    └────┬─────┘    └────┬─────┘
     │               │               │               │
     ▼               ▼               ▼               ▼
┌────┴─────┐    ┌────┴─────┐    ┌────┴─────┐    ┌────┴─────┐
│  Sensor  │    │   3D     │    │   Path   │    │ Actuator │
│ Visual.  │    │  Scene   │    │ Visual.  │    │ Visual.  │
└──────────┘    └──────────┘    └──────────┘    └──────────┘

Visualization Framework Architecture

┌─────────────────────────────────────────────┐
│              Application Layer               │
└───────────────────┬─────────────────────────┘
                    │
┌───────────────────▼─────────────────────────┐
│              Visualization Core              │
├─────────────┬─────────────┬─────────────────┤
│  Renderers  │  Layouts    │ Interaction     │
├─────────────┼─────────────┼─────────────────┤
│  2D         │  Timeline   │ Selection       │
│  3D         │  Split View │ Query           │
│  Plots      │  Multiview  │ Annotation      │
└─────────────┴─────────────┴─────────────────┘
                    │
┌───────────────────▼─────────────────────────┐
│              Data Adapters                   │
├─────────────┬─────────────┬─────────────────┤
│  Points     │  Images     │ Trajectories    │
│  Meshes     │  Video      │ Transforms      │
│  Volumes    │  Text       │ Semantic Data   │
└─────────────┴─────────────┴─────────────────┘

Multi-view Visualization Layout

┌───────────────┬───────────────┐
│               │               │
│  3D View      │  Raw Camera   │
│               │  Feed         │
│               │               │
├───────────────┼───────────────┤
│               │               │
│  LiDAR Top    │  System       │
│  View         │  State        │
│               │  Dashboard    │
└───────────────┴───────────────┘

Debugging Through Visualization

┌────────────────────────────────────────────────┐
│ Raw Data    Processed    Semantic    Decision  │
├────────┬────────┬─────────┬────────┬──────────┤
│        │        │         │        │          │
│Camera 1│Camera 2│ Detects │Tracking│ Planning │
│        │        │         │        │          │
├────────┴────────┴─────────┴────────┴──────────┤
│                                                │
│                 Timeline                       │
│  ◄─────────────────────────────────────────►  │
│                                                │
└────────────────────────────────────────────────┘

Temporal Data Visualization

     T=0         T=1         T=2         T=3    
┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
│    •     │ │          │ │          │ │    •     │
│   / \    │ │    •     │ │    •     │ │   / \    │
│  /   \   │ │   / \    │ │   / \    │ │  •   •   │
│ •     •  │ │  •   •   │ │  •   •   │ │ /     \  │
└──────────┘ └──────────┘ └──────────┘ └──────────┘
    State       Action      Prediction   Outcome

Robotic System Monitoring Dashboard

┌─────────────────────────────────────────┐
│ System Health                 ▲ CPU: 42% │
├──────────────┬──────────────┬───────────┤
│ 3D Map View  │ Camera View  │ Diagnostics│
│              │              │ ▓▓▓▓▓▓▒▒▒▒ │
│              │              │ ▓▓▓▓▓▓▒▒▒▒ │
│              │              │ Bat: 78%   │
├──────────────┴──────────────┼───────────┤
│ Console Output              │ Parameters │
│ [INFO] Localizing...        │ Max vel: 2m│
│ [INFO] Object detected      │ Map res: 5c│
│ [WARN] Low confidence       │ Conf th: .7│
└─────────────────────────────┴───────────┘

9. Drawing Practice - Key Diagrams for the Interview

Practice sketching these key diagrams that commonly appear in robotics visualization interviews:

  1. Camera projection model with distortion effects
  2. Point cloud registration before/after with correspondence lines
  3. SLAM factor graph with loop closure constraints
  4. Complete coordinate transformation chain from sensor to world
  5. Sensor fusion architecture with different fusion levels
  6. Multi-modal perception system with processing stages
  7. Visual-inertial odometry pipeline with data flow
  8. Uncertainty visualization in 3D (ellipsoids, particle sets)
  9. Robotics system architecture showing visualization components
  10. Timeline-based multi-sensor debugging interface