Why did the Robot Cross the Road?
summary
Build and result: a Tapsterbot delta robot controlled from Node.js using Johnny-Five and a new J5 Delta module to perform taps and swipes on a phone, a C920 manual-focus webcam controlled via a custom uvc-control Node package, and OpenCV template matching plus a Machina finite state machine to sequence actions. The bot taps every two seconds during play, watches 25‑second ads by opening the App Store and returning, and mostly works though lighting and focus make the live demo flaky. Tapsterbot delta robot template matching
AI-written summary · about
chapters (15)
- 0:00Introduction to the talk
- 1:52Problem with mobile games
- 3:16Introducing Tapster bot
- 5:02Delta robot kinematics
- 7:12Building J5 Delta library
- 8:16Webcam autofocus issues
- 10:02Reverse engineering webcam controls
- 11:07Computer vision with OpenCV
- 13:13Screen orientation correction
- 14:38Demo app setup
- 16:03Image recognition in OpenCV
- 18:52Finite state machine logic
- 21:18Crossy Road bot states
- 23:46Live demo attempt
- 27:26Conclusion and Q&A
related talks
39mRise of the Robots
29mLearnBot, the open source robot for education
15mRobots at your fingerprints
21mSo you want to build a robot?
15mRobot Hand Driven by Computer Vision and Reinforcement Learning
30mMuch Robots Very Java So IoT
11mI built a robot to cheat at Pokemon!!
57mThe Convergence of Robotics, the Web, and IoT