ProjekinAja. Robotics Mastery 0 / 56 done
ZERO → HERO · THE COMPLETE DISCIPLINE

Robotics
Mastery

From coordinate frames and rigid-body transforms to deploying a real autonomous person-following robot — taught through FollowBot, a complete mobile manipulator you build, tune, and harden across all six stages. Failure modes before happy paths; mechanisms over incantations.

6Stages
56Lessons
56Widgets
FollowBotthread
0 of 56 lessons complete
1

Foundations: Sense, Think, Act, and Compute

9 lessons

Before any algorithm runs, a robot must be defined — as a sense-think-act loop, as a collection of rigid bodies with coordinate frames, as a ROS 2 computation graph, and as a set of imperfect sensors and actuators. This stage installs four non-negotiable mental models: the closed-loop embodied agent, the SE(3) transform as the grammar of robotics, the ROS 2 communication architecture, and sensor noise as the central engineering problem.

FollowBot milestone: FollowBot's full system architecture is defined: URDF built from scratch with correct inertial parameters, TF tree complete (map → odom → base_link → laser → camera), sensor topics publishing simulated data in Gazebo, first custom ROS 2 node running, requirements specification constraining all downstream design decisions.

2

Kinematics and Velocity: The Geometry of Motion

9 lessons

Kinematics answers the question: given joint angles or wheel velocities, where does the robot end up, and how fast does it move? This stage establishes the two kinematic models FollowBot depends on: the differential-drive mobile base and the serial manipulator arm. The Jacobian is introduced as the spine of the course — connecting configuration space to task space, velocity to force, and singularities to catastrophic failure.

FollowBot milestone: Diff-drive kinematic model implemented in Python/ROS 2: forward model (wheel velocities → robot velocity), inverse model (desired cmd_vel → wheel commands), odometry integration producing /odom. Arm FK and Jacobian derived from DH parameters; singularity monitor live; workspace visualization showing FollowBot's reachable space.

✓
LESSON 2.1
Differential Drive Kinematics: FollowBot's Motion Model
The differential-drive model is elegant in theory and wrong in practice — knowing exactly how and why…
✓
LESSON 2.2
Wheel Odometry: Integration, Drift, and Dead Reckoning
Odometry is always wrong — the question is how wrong, how fast, and whether your estimator accounts…
✓
LESSON 2.3
Forward Kinematics and Denavit-Hartenberg Parameters
DH parameters are a compact notation — but they hide enough subtleties that a single sign error silently…
✓
LESSON 2.4
The Jacobian: Velocity Kinematics and the Spine of Robotics
The Jacobian is not just 'the matrix of partial derivatives' — it is the object that connects geometry…
✓
LESSON 2.5
Singularity Analysis and Manipulability
Singularities are not edge cases — every arm has them and any controller that ignores them will eventually…
✓
LESSON 2.6
Inverse Kinematics: Closed-Form and Numerical Methods
IK has no general closed form — and the numerical methods that substitute for it are local, fragile,…
✓
LESSON 2.7
Velocity Control in Practice: cmd_vel to Wheel Commands
Publishing cmd_vel is not controlling a robot — understanding the inner velocity control loop is what…
✓
LESSON 2.8
Trajectory Parameterization: Profiles, Splines, and Jerk
A trajectory that ignores jerk limits will make the robot vibrate, excite resonances, and shake sensors…
✓
LESSON 2.9
Mobile Manipulator Kinematics: When FollowBot Reaches Out
Stacking a mobile base and an arm creates redundancy — and redundancy without a null-space policy creates…
3

Dynamics and Control: Making Things Move Right

9 lessons

Kinematics tells you where the robot is; dynamics tells you how to make it go there. This stage derives the equation of motion from both Newton-Euler and Lagrangian formulations, then builds controllers in pedagogical order: stability first, then transient performance, then optimality. PID is derived from its Laplace transform. LQR is built from the Riccati equation. The sim-to-real gap between the dynamic model and real hardware is introduced here — not deferred to Stage 6.

FollowBot milestone: Full dynamics model for FollowBot's arm derived; PID velocity controller tuned to a quantitative settling-time spec using Ziegler-Nichols then refined via root locus; feedforward from inverse dynamics reducing tracking error; figure-eight trajectory tracked with measured cross-track error below spec.

✓
LESSON 3.1
Newton-Euler Dynamics: Forces, Torques, and Rigid Bodies
Every dynamic model starts with F = ma — the art is accounting for all the forces your model ignores.
✓
LESSON 3.2
Lagrangian Dynamics: Energy Methods and the Equation of Motion
The Lagrangian is not just an alternative derivation — it is the formulation that makes the structure…
✓
LESSON 3.3
Actuator Models and the Sim-to-Real Dynamics Gap
The equation of motion assumes ideal actuators — real motors saturate, have backlash, and respond to…
✓
LESSON 3.4
PID Control: Derivation, Tuning, and When It Fails
PID is the most deployed controller in history and the most mistuned — derive it from the Laplace…
✓
LESSON 3.5
Stability: Lyapunov Theory and Phase Portraits
Asserting 'the system is stable' without a proof is not engineering — Lyapunov's method gives you the…
✓
LESSON 3.6
Frequency-Domain Analysis: Bode Plots and Robustness Margins
Gain margin and phase margin are not performance metrics — they are the quantitative answer to 'how wrong…
✓
LESSON 3.7
Feedforward and Computed-Torque Control
Feedforward from the inverse dynamics model reduces tracking error by orders of magnitude — but only if…
✓
LESSON 3.8
State-Space Representation and LQR
LQR is PID with a principled cost function — understand the Riccati equation and you can tune the Q/R…
✓
LESSON 3.9
Model Predictive Control: Constraints as the Killer Feature
MPC beats PID when you have hard constraints on states or inputs — without constraints it is usually not…
4

State Estimation: Knowing Where You Are

9 lessons

State estimation is the bridge between noisy sensors and confident beliefs. A robot that does not know where it is cannot plan. This stage derives the Bayes filter from first principles, builds the Kalman filter as the MMSE solution for linear Gaussian systems, extends to the EKF for nonlinear systems, and introduces the particle filter for multimodal beliefs. Filter consistency tested via NEES and NIS is taught as a mandatory engineering practice.

FollowBot milestone: Full EKF operational on FollowBot: state vector [x, y, theta, v, omega], fusing wheel odometry + IMU + LIDAR scan-matching updates. Filter consistency validated with NEES test on recorded data. Robot localizes in a known map with position error below 0.1 m RMS. Sensor dropout behavior characterized and documented.

✓
LESSON 4.1
Probability for Robotics: Belief, Bayes, and Why Point Estimates Fail
A robot's state is a probability distribution, not a number — and every algorithm that pretends otherwise…
✓
LESSON 4.2
The Bayes Filter and the Predict-Update Cycle
Every state estimator from KF to particle filter is a Bayes filter with different assumptions — understand…
✓
LESSON 4.3
The Kalman Filter: Derivation and Geometric Intuition
The Kalman gain is not a magic formula — it is the minimum-variance solution to a linear regression…
✓
LESSON 4.4
Kalman Filter Tuning: Q, R, and the Art of Honest Uncertainty
Q is not process noise — it is your honest admission of how much you do not know about your model; getting…
✓
LESSON 4.5
The Extended Kalman Filter: Nonlinear Systems and Jacobian Linearization
The EKF's Jacobian linearization is a first-order approximation — understand exactly when it fails and…
✓
LESSON 4.6
The Unscented Kalman Filter: When the Jacobian Is Not Enough
The UKF propagates a set of carefully chosen sigma points instead of linearizing — it costs more but…
✓
LESSON 4.7
Particle Filters: Non-Gaussian and Multimodal Beliefs
The particle filter is the only estimator that handles genuinely multimodal beliefs — and the kidnapped…
✓
LESSON 4.8
IMU Preintegration and High-Rate Sensor Fusion
Running an EKF at IMU rate (200 Hz) with a 10 Hz LIDAR is architecturally wrong — IMU preintegration…
✓
LESSON 4.9
Multi-Sensor Fusion on FollowBot: Architecture and Covariance Design
The multi-sensor EKF is only as good as its weakest covariance matrix — and most practitioners set those…
5

Perception and SLAM: Seeing and Mapping

10 lessons

Perception transforms raw sensor data into the structured representations that estimation, planning, and control can use. This stage starts with the geometric foundation of cameras, moves through LIDAR point-cloud processing, covers object detection and tracking for person-following, and culminates in SLAM. Every algorithm is taught with its failure mode as co-equal to its nominal behavior.

FollowBot milestone: Camera intrinsics calibrated (reprojection error < 0.5 px); LIDAR-camera extrinsic calibrated with validated residual metric; YOLOv8 person detection integrated in ROS 2; LIDAR cluster + camera bbox fusion producing stable 3D person position at 10 Hz; SLAM Toolbox building an occupancy map of the test environment; occupancy grid inflated and ready for navigation.

✓
LESSON 5.1
Camera Geometry: The Pinhole Model, Distortion, and Why Calibration Is Not Optional
A camera that has never been calibrated is a measurement device that has never been zeroed — every pixel…
✓
LESSON 5.2
Stereo Vision and RGB-D Depth Sensing
Depth uncertainty grows quadratically with distance for stereo — knowing this quantitatively tells you…
✓
LESSON 5.3
LIDAR Point Cloud Processing: From Scan to Structure
A raw LIDAR scan is not a map — voxelization, outlier removal, and ground segmentation must run before…
✓
LESSON 5.4
Feature Detection and Visual Odometry
Visual odometry gives you metric ego-motion from a camera — but scale ambiguity in monocular VO means…
✓
LESSON 5.5
Person Detection and 3D Localization: LIDAR + Camera Fusion
A bounding box tells you what — LIDAR tells you where — and getting the coordinate transform wrong between…
✓
LESSON 5.6
Object Tracking: Kalman + Hungarian and the ID-Switch Problem
The ID-switch failure — when your tracker assigns the wrong ID to a returning person — is not an edge…
✓
LESSON 5.7
Occupancy Grid Mapping: The Map Representation That Works
The log-odds representation is not a design choice — it is numerical necessity; linear probability updates…
✓
LESSON 5.8
Scan Matching and ICP: LIDAR-Based Pose Estimation
ICP is a local optimizer — give it a bad initial guess and it will converge confidently to the wrong…
✓
LESSON 5.9
SLAM: Simultaneous Localization and Mapping
SLAM is the hardest single problem in mobile robotics — and false loop closures are the single most…
✓
LESSON 5.10
Production SLAM Systems: Cartographer, SLAM Toolbox, and Choosing Wisely
The choice of SLAM system is not a personal preference — it is a constraint satisfaction problem between…
6

Planning, Navigation, and Deployment

10 lessons

Planning answers the question: given that the robot knows where it is and what it sees, how does it decide what to do? This final stage covers configuration space, global and local planners, the Nav2 stack, behavior trees, and the sim-to-real gap as a continuous thread. Sim-to-real failures are introduced by category and diagnosed systematically. The stage ends with a full-system capstone.

FollowBot milestone: Full Nav2 stack configured: A* global planner on SLAM map, DWA local planner with FollowBot's dynamic constraints, costmap with LIDAR obstacle layer and inflation. Behavior tree orchestrating: detect person → navigate to tracked goal → maintain following distance → handle occlusion → recover. Sim-to-real checklist signed off. Capstone: documented system architecture with honest assessment of each subsystem's real-world performance vs. simulation.

✓
LESSON 6.1
Configuration Space: The Correct Abstraction for Planning
Every planner operates in configuration space whether you know it or not — misunderstanding C-space produces…
✓
LESSON 6.2
Graph Search: A*, Heuristics, and Costmap Planning
A* is only as good as its heuristic — an inadmissible heuristic produces fast plans that are not optimal…
✓
LESSON 6.3
Sampling-Based Planning: RRT and RRT*
RRT finds a feasible path fast but the path is usually ugly — RRT* fixes this asymptotically, and both…
✓
LESSON 6.4
Local Planners: DWA, TEB, and Real-Time Obstacle Avoidance
The local planner is where theory meets friction — DWA samples velocity space and TEB optimizes an elastic…
✓
LESSON 6.5
The Nav2 Stack: Architecture, Lifecycle Nodes, and Failure Modes
Nav2 is an ecosystem of composable plugins — treat it as a black box and you will tune parameters blindly…
✓
LESSON 6.6
Behavior Trees: Architecture, Ticks, and Reactive Mission Management
A bad state machine cannot recover from unexpected events — a behavior tree can, but only if you design…
✓
LESSON 6.7
The Sim-to-Real Gap: Taxonomy, Root Causes, and Honest Mitigations
The sim-to-real gap has four distinct failure classes — conflating them leads to applying the wrong…
✓
LESSON 6.8
Hardware Bring-Up, Calibration, and Field Diagnostics
Hardware bring-up without a protocol is gambling — a systematic sequence (power → comms → sensors →…
✓
LESSON 6.9
Safety Architecture: Watchdogs, E-Stops, and Speed Governors
Safety is not a feature you add at the end — a robot that cannot fail safely is not safe, regardless…
✓
LESSON 6.10
Capstone: FollowBot Complete — Architecture, Gaps, and What Comes Next
The gap between a robot that works in the lab and one that can be trusted in the field is not an algorithm…
🎓

Certificate

Finish all 56 lessons to unlock

Complete every lesson across all six stages — from coordinate frames to full-system deployment — and claim your personalized certificate of completion. FollowBot's journey, your credentials.

📚

Sources & Further Reading

Canonical works behind this course

These are the real, canonical works behind this course — go deeper here. References are grouped by the stage they most directly underpin.

Foundations — Rigid-Body Geometry, ROS 2, and Sensors

Kinematics — Forward, Inverse, and the Jacobian

Dynamics and Control — PID, LQR, MPC

State Estimation — Bayes, Kalman, EKF, Particle Filters

Perception and SLAM — Vision, LIDAR, and Mapping

Planning, Navigation, and Deployment