The full course, module by module

A grounding in what makes a robot "autonomous": the sense–plan–act loop, and the math you'll lean on for the rest of the course.

• What is a robot? Sense–Plan–Act loop
• Coordinate frames & transforms (2D/3D)
• Linear algebra refresher: vectors, rotation matrices
• Probability basics for sensor noise
• Where AI fits: perception vs. planning vs. control
• Course roadmap & the capstone robot preview

The everyday tooling every ROS2 developer needs before writing a single node.

• Ubuntu & shell essentials (bash, permissions, processes)
• Git & GitHub workflow for robotics projects
• Python for robotics: classes, threading basics, NumPy
• Virtual environments & dependency management
• SSH & remote development on a Raspberry Pi
• Debugging tools: htop, journalctl, tmux

The ROS2 mental model: how independent nodes talk to each other, and the tools to inspect that traffic live.

• Nodes, topics, publishers & subscribers
• Services vs. actions — when to use each
• Packages, workspaces & colcon build
• Launch files & parameters (YAML)
• tf2: transforms between frames
• rviz2 & rqt for visualization and debugging
• Writing your first Python (rclpy) node

How two independently driven wheels turn into a robot's linear and angular velocity — and how to keep track of where the robot actually is.

• Differential drive kinematics: v, ω from wheel speeds
• Wheel encoders & odometry integration
• IMU basics & sensor fusion (complementary/Kalman intro)
• PID control for wheel velocity
• Publishing /odom and the odom → base_link transform
• Common drivetrain mistakes and how to avoid them

The sensor suite that turns a blind chassis into a robot that understands its surroundings.

• 2D LiDAR: sensor_msgs/LaserScan
• Cameras: intrinsics, calibration, image topics
• Ultrasonic & IR proximity sensors
• IMU: accelerometer, gyroscope, magnetometer
• Sensor fusion basics for a mobile base
• Noise, drift, and how to sanity-check sensor data

From classic image processing to real-time neural object detectors running on-device.

• OpenCV fundamentals: filtering, contours, color thresholds
• CNNs in 20 minutes: convolution, pooling, feature maps
• Object detection models: YOLO family, SSD, MobileNet-SSD
• Evaluating detectors: IoU, precision/recall, NMS
• Model optimization: quantization, ONNX, TensorRT
• Deploying a detector on Raspberry Pi / Jetson inside ROS2
• Publishing detections as vision_msgs

How Nav2 turns a map and a goal pose into a safe path across the environment.

• Nav2 architecture overview: servers & lifecycle nodes
• Costmap2D: static, obstacle, and inflation layers
• Global planners: Dijkstra, A*, NavFn, Smac Planner
• Heuristics & grid vs. graph-based planning
• Global vs. local planning — why you need both
• Configuring planner plugins via YAML

Turning a global path into safe, real-time velocity commands that react to obstacles the map didn't know about.

• Local planners: DWA (Dynamic Window Approach), TEB
• Trajectory scoring: obstacle cost, path alignment, speed
• Controllers: Regulated Pure Pursuit, velocity smoothing
• Publishing /cmd_vel and closing the control loop
• Recovery behaviors: spin, back-up, clear costmap
• Tuning for a small differential-drive base

Building a map from scratch, and figuring out where your robot is inside one it already has.

• Mapping vs. localization — the two halves of SLAM
• slam_toolbox for 2D LiDAR mapping
• AMCL: particle-filter localization on a known map
• Saving & loading maps (map_server)
• Loop closure & drift correction, briefly
• When to remap vs. relocalize

How Nav2 sequences planning, recovery, and detection into one coherent autonomous behavior.

• BT Navigator & behavior tree basics
• Composing custom behavior trees (XML)
• Multi-node system architecture for a full robot stack
• Combining navigation with object detection (e.g. "patrol & report")
• Logging, diagnostics, and ros2 bag for debugging

Everything from modules 1–10, now applied to real hardware costing under ₹50,000.

• Chassis assembly & motor wiring
• Flashing & configuring the Raspberry Pi
• Bringing up ROS2 drivers for motors, LiDAR & camera
• Calibrating odometry on real hardware
• Running Nav2 + SLAM on the physical robot
• Safety: e-stops, current limits, battery care

Bring every module together into one demo: a robot that maps a room, localizes itself, navigates to waypoints, and reports what it sees.

• Define a mission: patrol route + object-of-interest detection
• Integrate SLAM, Nav2, and your object detector
• Handle edge cases: dynamic obstacles, lost localization
• Record a demo run with ros2 bag
• Present & document your build