Mobile Robotics Curriculum

Table of Contents

• Overview
• Pre-Stage-2 Warm-up: XRP
• Scope
• Candidate Hardware
• Planned Modules
• References

Overview

🚗 Stage 2 of a staged progression toward autonomous quadcopter flight, following the SO-101 manipulator curriculum (Stage 1). Hardware is now decided and the module structure is drafted (see Candidate Hardware and Planned Modules below, and the companion buildout doc) — remaining work is verifying exact BOM pricing/stock immediately before ordering and filling in per-module reading materials with the same rigor as Stage 1’s CMU-course-substitute research.

Pre-Stage-2 Warm-up: XRP

✅ In hand — ordered and received. Before committing to Stage 2’s own hardware decision (below), warm up on odometry, encoders, and proportional control — the cheapest, lowest-risk fundamentals in this stage’s Scope — using WPI’s XRP (Experiential Robotics Platform).

Hardware: XRP Kit — $119.95 from SparkFun. Controller is a Raspberry Pi RP2350B (dual Cortex-M33, 16MB flash, 8MB PSRAM) with an onboard 6-DoF IMU (LSM6DSOX), two Qwiic connectors, and dual-channel motor drivers for up to four motors. The kit itself ships with 2 encoded drive motors + 2 casters (differential drive, not a balancing platform — relevant only to this stage, not Stage 3), an ultrasonic rangefinder, a line-following sensor, a servo, and a 3D-printed chassis. Batteries and USB-C cable are not included. Programmed with WPILib — the same toolchain used for full-size FIRST Robotics Competition (FRC) robots — via Blockly or Python.

Learning materials — two curricula exist, pick based on self-study fit: - Introduction to Robotics (WPI Global STEM Education Initiative, ReadTheDocs) — preferred primary path, since it’s structured for individual self-paced learners rather than a classroom. Modules run Introduction → Driving → Sensors → Manipulator (a small robot arm attachment — direct conceptual overlap with SO-101 skills, just at toy scale) → Capstone (an “autonomous delivery robot” challenge). Each module is a sequence of ordered, interactive challenges on the physical robot. - WPI XRP Curriculum (wp.wpi.edu) — the original K-12 classroom curriculum, four units with full lesson-plan PDFs (slides, guided notes, homework, answer keys, grading rubrics): Unit 1 (assembly, Blockly/Python basics, Gate Maze Challenge), Unit 2 (ultrasonic sensing, proportional control, encoders, Moving Maze), Unit 3 (line following — reflectance + ultrasonic, on/off vs. proportional control), Unit 4 (servo integration, final time-trial capstone). Useful as a reference for classroom-tested exercise design even if not followed lesson-by-lesson. - WPILib XRP docs — programming reference: hardware & imaging, getting to know your XRP, hardware support, web UI, programming the XRP. - experiential.bot — community hub: forum, helpdesk, and 3D-printable hardware extensions (XBS build system, legacy parts).

Scope

Core topics this stage needs to cover, at minimum: - Odometry (wheel encoders, dead reckoning, and its drift/error accumulation) - Localization and mapping (SLAM — at least a working intuition for EKF-SLAM and/or a modern LiDAR/visual SLAM pipeline) - Path planning (A, RRT/RRT, or similar) and obstacle avoidance - Sensor fusion (combining odometry, IMU, and exteroceptive sensors — LiDAR or depth camera) - A middleware layer (most likely ROS2, given it’s the de facto standard and will also be relevant for the quadcopter-autonomy stage)

Candidate Hardware

Decision made — see Mobile Robot Buildout doc for the full comparison and BOM. Going with a DIY ROS2 rover, following the free Articulated Robotics tutorial series (Josh Newans): source parts yourself (Raspberry Pi 4B/5, RPLiDAR A1, Arduino Nano for motor control, brushed DC gearmotors with encoders, camera), assemble, and write the ros2_control hardware interface from scratch — same hands-on-first reasoning as the SO-101’s Path C decision. Estimated cost ~$275–350, versus $900+ for either official TurtleBot platform.

Two other directions were weighed and ruled out (detail in the buildout doc): - LeKiwi — LeRobot-native mobile base, ~$300–500. Stays inside the LeRobot/Feetech ecosystem already invested in, but its tooling is manipulation-focused, not SLAM/path-planning focused — the actual Stage 2 content would need to be layered on top rather than coming for free. - TurtleBot 3 / TurtleBot 4 — official ROS2 reference platforms with excellent documentation, but $900–$1,900 and largely pre-integrated, which removes exactly the hardware-integration learning value this stage is for.

Planned Modules

Derived directly from the Articulated Robotics tutorial series structure (see Curriculum Mapping in the buildout doc for the full diagram):

• Concept & Design — URDF modeling of the rover, Gazebo simulation (zero-hardware-cost first pass, same “simulate first” principle as the quadcopter stage’s Phase 4)
• Hardware Bring-up — Raspberry Pi OS/ROS2 setup, power system design, LiDAR integration (RPLiDAR A1 → LaserScan messages), camera integration
• ros2_control — hardware interface concepts, then wiring it to the real robot (encoders → odometry)
• Teleoperation — manual driving via ROS2 topics, sanity-checks the full hardware stack before autonomy
• SLAM — slam_toolbox against the RPLiDAR A1, building a map of a real space
• Nav2 — path planning and autonomous navigation using the map from the SLAM module
• Object Tracking — OpenCV-based detection/following using the camera, a stretch module tying back into Stage 1’s computer-vision skills

Milestone: the rover builds and holds a live SLAM map of a real room, then autonomously navigates to a commanded goal pose via Nav2 — directly analogous to the SO-101’s teleoperation-to-autonomous-policy arc, but for navigation instead of manipulation.

References