This repository contains the autonomous driving code and mission logic developed for the 3rd Future Automotive Autonomous Driving Competition, hosted in 2024.
The challenge involved completing multiple mission stages using a limited set of sensors under strict environmental and timing conditions.
🥉 Bronze Prize Winner – 3rd Future Automotive Autonomous Driving Competition (2024)
📸 Featured in Dankook University News:
Click to read full article (Korean)
"Even though we didn't modify the motor output like other teams did, we safely completed every mission with a stable system. That’s something we’re truly proud of."
— From the team interview in the DKU article
This competition limited the type and number of sensors available for all teams.
- Allowed Sensors:
- 1 × LiDAR
- 2 × Cameras
-
Basic Driving Mission
→ Top 10 teams with fastest lap times (2 laps) qualify for the next round. -
Main Missions (combined score ranking):
- Obstacle Avoidance
- Traffic Light Response
- Autonomous Parking
- 🧠 Team Lead
- 🧭 Obstacle Avoidance Logic
🅿️ Autonomous Parking Algorithm- 🧾 Data Collection & YOLOv8 Labeling for Lane Detection
Due to limited sensors, traditional lane detection was not viable.
Instead, we:
- Recorded a bag file of the track
- Annotated images manually
- Trained a YOLOv8 segmentation model to detect lane regions
➡️ As training data increased, lane tracking stability improved significantly.
- Defined a forward-facing ROI on LiDAR data
- Implemented smooth avoidance behavior upon detecting objects within a threshold range
- Used YOLOv8 object detection
- Detected red/green states and triggered corresponding vehicle actions (stop/go)
- Used a right-side ROI window to detect free space
- Performed hardcoded parking maneuvers based on vehicle position and spacing
(Starts at 59:56)
(Obstacle, Traffic Light, Parking)
(Starts at 1:57:47)
Despite long preparation, we were limited by sensor count and hardware power.
- Our vehicle was relatively light but the motor performance lagged compared to other teams.
- In the first stage, we barely qualified due to speed limitations.
- In the final stage, we successfully completed all missions including obstacle avoidance, traffic light handling, and parking.
However, we only received the Bronze Prize.
After the competition, we found out many teams had force-overridden motor output limits — something we consciously avoided due to safety and competition fairness.
While it was disappointing, we take pride in building one of the most reliable and safest autonomous systems in the competition.