Building a self driving RC car
summary
A team converts a €200 RC car with a €150 sensor budget into an autonomous racer using Raspberry Pi, OpenCV and Java on Vert.x, controlling servos via servoBlaster (echo commands). Four teams race: a 50m straight course plus stop zone; two cars complete the autonomous run and one wins by ~2 meters. Image pipeline uses Canny edges, a probabilistic Hough transform, ROI cropping and angle-based steering; safety uses a 100 ms heartbeat. Major pain points: lighting, floor spots, oversteer and most processing running on a laptop rather than the Pi.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 2:29How did this all start
- 4:40The kit
- 12:25The interface
- 12:55The challenge
- 14:10Image processing pipeline
- 15:45Code
- 19:35Architecture
- 23:13Testing
- 28:53Would you drive it
- 29:16Self driving in the real world
- 31:35Waymo car
- 33:54Coma AI
- 35:04Safety precautions
- 36:19Ethical dilemma
- 36:58Cats vs dogs
- 39:23Who to preserve
- 43:26Questions
- 48:06Wrap up
related talks
35mGetting started with a self-driving RC car
23mRacing Robocars
15mFrom the Racetrack to the Road - How Open Source Boosts Autonomous Vehicle Development
31mAutonomous driving, how Python drives Formula Racing cars
01h 51mLearn to Drive (and Race!) Autonomous Vehicles
47mAchieving Software Quality Excellence Thru Software-Hardware Co-Design & Co-Verification
45mTutorial: Building a Drone, From Scratch
55mUsing a Lidar for Robot Navigation in a Room