A robot that worked once is a good start. A robot that works 19 times out of 20 is a finished project. The difference is testing.
Key idea
Engineers work in a cycle: design, build, test, change one thing, test again. Run each version several times, record every run in a test log, and compare success rates.
| Version | Change | Runs | Successes | Rate |
|---|---|---|---|---|
| 1 | first build | 10 | 6 | 60% |
| 2 | slower turns | 10 | 8 | 80% |
| 3 | new tires | 10 | 7 | 70% |
Why it works
If you change three things at once and it gets better, you cannot tell which change helped, or whether one hurt. With only a few runs, luck can look like progress; more runs make the comparison fair. Competitions like FIRST reward teams that test this way.
Worked example: Success rate
17 successes in 20 runs.
- $17/20 = 0.85$.
Answer: 85%.
Common mistake
Changing three things at once, or only writing down the runs that worked.
A project ladder for your portfolio
- Rung 1: a line follower with two light sensors and a state machine.
- Rung 2: an obstacle avoider with an ultrasonic sensor and encoder odometry; log its square-test drift.
- Rung 3: a two-joint servo arm that reaches typed-in targets with inverse kinematics.
- Rung 4: a wheel-speed PID controller with a plot of step responses at three gains.
- Rung 5: a ball chaser with a camera and color thresholding.
- Rung 6: the same robot in ROS 2: nodes for the camera, the controller and the motors, tested first in simulation.
For each rung, write a one-page report: the goal, a photo or short video, the code, the test log, and what you would change next. That stack of pages is a portfolio.