Portfolio of things that moved

27 practical projects

A robotics portfolio is made of things that moved, not courses you finished. Check them off as you film, write up, and push them to GitHub.

1. Month 1

   Open module

   • Project 01

     build a voltage divider in Falstad, calculate the output voltage by hand, then confirm it in the simulator. Then build a transistor switch that turns an LED on from a logic-level input, since that is the exact circuit that will later let a 3.3V microcontroller pin control something that needs more current than the pin can supplyNot completed

   • Project 02

     buy a kit and a multimeter, then measure things. Measure the voltage of a battery, the resistance of five random resistors and check them against the colour bands, then use continuity mode to find a deliberately broken wire you make yourself. This sounds trivial and it is the single skill that will save you the most hours over the next six monthsNot completed

   • Project 03

     solder header pins onto a cheap breakout board, then test every pin with continuity mode, and do the whole thing three times. Then desolder one and resolder it, because removing components badly is how most beginners destroy boardsNot completed

   • Project 04

     from today, every project you build lives in a GitHub repo with a README that has a photo, a wiring description and a one-paragraph explanation of what broke and how you fixed it. That last part is what makes a repo look like engineering instead of a tutorialNot completed

2. Month 2

   Open module

   • Project 01

     build a reaction-timer game. An LED fires after a random delay, a button stops the clock, and the time in milliseconds prints to serial. It uses interrupts, debouncing and non-blocking timing, and it has a score, which makes it demonstrable in fifteen seconds of videoNot completed

   • Project 02

     build something that could not exist on an Arduino Uno. Serve a web page from the ESP32 that shows live sensor readings and has buttons that drive a servo, then access it from your phone on the same network. This teaches WiFi, HTTP handling and asynchronous work at the same timeNot completed

   • Project 03

     drive one DC motor forward and backward at five different speeds using PWM, then add an encoder and write a function that turns the wheel exactly one full revolution regardless of battery voltage. The second half is your first real closed loop, and it is much harder than it soundsNot completed

   • Project 04

     mount an IMU on a board, print the pitch angle to serial, hold the board perfectly still and watch the number drift anyway. Now add a complementary filter and watch that drift disappear, which is the single most important lesson in state estimation and it took you twenty minutesNot completed

   • Project 05

     Practice task for the month: build the line follower first, tune it with a P controller, then add the D term and watch the oscillation disappear, filming both versions. The video of a badly tuned robot next to the same robot tuned properly is one of the most persuasive things a beginner can put in a portfolio, because it proves you understand the loop rather than having copied a gainNot completed

3. Month 3

   Open module

   • Project 01

     model a bracket that holds the exact servo you bought, with correctly sized screw holes and a shaft clearance, from the manufacturer's datasheet drawing rather than by eye. Then print it and see if it fits, which it probably will not the first time, and that failure is the lessonNot completed

   • Project 02

     print the tolerance gauge, write down your machine's actual clearance numbers, then design and print a two-part snap-fit enclosure for your ESP32 that closes without glue. Iterate until it clicks properly. This is the loop that all mechanical design is made ofNot completed

   • Project 03

     design and print a simple planetary or cycloidal reducer for a NEMA17 stepper or a hobby motor, and accept that the first one will be bad. Measure the backlash by holding the output and rocking it, then redesign to reduce it. Reference builds are on Instructables and Hackaday if you want a starting geometryNot completed

   • Project 04

     build the SO-101, calibrate every servo, and teleoperate the follower with the leader. Then design and print your own gripper fingers to replace the stock ones, in TPU, and test them on three objects of different shapes. The arm is the platform for months five and six, and the custom fingers are the part that proves you can design as well as assembleNot completed

4. Month 4

   Open module

   • Project 01

     build a multi-node system with no robot and no simulator at all. A sensor publisher, a processing node, a service-based configuration node and a parameterised aggregator, with your own custom .msg and .srv, wired together in a Python launch file that takes arguments. Record it with ros2 bag and replay it. This costs nothing and it proves you understand the graph rather than that you can run turtlesimNot completed

   • Project 02

     design your own robot in xacro, not TurtleBot. A differential-drive base, a sensor mast and a two-DOF pan-tilt head, with correct inertias and separate collision and visual geometry. Drive the joints with joint_state_publisher_gui and view the full TF tree in RViz. A screenshot of your own robot in RViz is the most legible signal in a beginner portfolioNot completed

   • Project 03

     put your xacro robot from the last section into Gazebo, add a lidar plugin and a camera plugin, and build a custom SDF world with obstacles for it to sit in. Confirm the sensor data appears on ROS 2 topics and renders in RVizNot completed

   • Project 04

     add <ros2_control> tags to your robot, configure diff_drive_controller and joint_state_broadcaster through YAML, and drive it in Gazebo with keyboard teleop. Then write an action server that drives a commanded distance and reports progress, with feedback and cancellation. Actions plus ros2_control in one project is a genuinely strong portfolio pieceNot completed

   • Project 05

     run SLAM Toolbox in your Gazebo world, teleop around it, save the map, then switch to localization mode and send Nav2 goals from RViz. Tune the costmaps until it stops clipping corners. Then screen-record it. A video of a robot autonomously navigating a map it built itself is the most compelling artifact a self-taught roboticist can produceNot completed

5. Month 5

   Open module

   • Project 01

     take the balancing robot from month two and implement three controllers on the same hardware: P only, PD, then PID with feedforward. Log the step response of each to a CSV, plot all three, and write up which one you would ship and why. That plot is worth more in an interview than any certificateNot completed

   • Project 02

     implement LQR for a simulated cart-pole in Python, then implement the same thing with a hand-tuned PID and compare how each handles a disturbance. Tedrake's notes give you the model, so you are implementing rather than derivingNot completed

   • Project 03

     compute the forward kinematics of your SO-101 arm by hand from its link lengths, then verify against the Robotics Toolbox. Then write a numerical IK solver that moves the end effector to a commanded XYZ, and watch what it does near a singularity. Feeling the arm lose a degree of freedom is what makes the concept stickNot completed

   • Project 04

     calibrate an actual camera with a printed chessboard, save the intrinsics, then write a script that detects a coloured object and estimates its position in 3D relative to the camera. Then move the lighting and watch it fail, and fix it. That failure-and-fix write-up is portfolio materialNot completed

   • Project 05

     get MoveIt 2 planning motions for a simulated arm, add collision objects to the planning scene, and execute a pick and place. Then make it fail by placing an obstacle in the only viable path and observe how the planner behaves. Understanding planner failure is more valuable than watching it succeedNot completed

6. Month 6

   Open module

   • Project 01

     record 50 demonstrations of a single simple task on your SO-101, such as picking up a cube and dropping it in a bin, train an ACT policy, and deploy it. It will work maybe half the time. Then record 50 more demonstrations covering the failure cases and retrain. Document the success rate before and after. That number, and the fact that you measured it, is the portfolio pieceNot completed

   • Project 02

     train a quadruped locomotion policy in MuJoCo Playground from one of the Colab tutorials, then change the reward function and observe how the gait changes. You do not need hardware and you do not need a GPU beyond what Colab gives you freeNot completed

   • Project 03

     take your three best projects and rewrite their READMEs. Each one needs a video at the top, a wiring or architecture diagram, the actual numbers you measured, and a section titled "what broke and how I fixed it". That last section is the single highest-value thing in a self-taught portfolio, because it is the part that cannot be faked from a tutorialNot completed

   • Project 04

     have someone interrogate you about your own repo for twenty minutes. Not the concepts, your specific code. Why that gain, why that sensor, what happens if the battery sags, what did you try before this worked. If you cannot answer three levels deep, the project is not ready to be on your CVNot completed