We are from South Korea to win this HACKATHON !!!✊✊✊
| Name | GitHub Handle | Role(s) |
|---|---|---|
| Juhun Kang | Ju-Daeng-E | Team Leader/Developer |
| DongMin Jang | jjangddung | Developer |
| JaeHong Lim | Jae Hong Lim | Developer |
| MinHyeok Choi | kevin choi | Developer |
| JeongMin Jo | JeongMin | Developer |
SDV Lab Challenge
Current vehicle pursuit scenarios are dangerous and inefficient:
- Driver overload: Officers must drive at high speeds while tracking suspects, using radio, and reporting
- Poor coordination: Other units only hear audio descriptions, can't visualize suspect location
- No video sharing: Command center can't see real-time pursuit footage
- Manual tracking: Officers struggle to keep visual contact during pursuit
1. Automatic Lock-on
- Semantic LiDAR detects collision/crime events
- Automatically identifies and locks onto suspect vehicle
- ObjIdx-based tracking maintains lock across frames
- No manual targeting needed
2. Autonomous Pursuit
- Chase vehicle automatically follows locked target
- Officer focuses on assessment and decision-making, not driving
- Safer pursuit with AI-controlled vehicle dynamics
3. Real-time Video Streaming
- Live pursuit camera feed via Zenoh → WebSocket
- Command center monitors situation on web dashboard
- Better tactical decisions with visual context
4. VSS-based Coordinate Broadcast
- Suspect GPS coordinates in standardized VSS format
- Real-time broadcast via Zenoh to all nearby units
- 50ms update rate for accurate positioning
5. Coordinated Encirclement
- Encirclement units receive target coordinates automatically
- A* pathfinding calculates optimal interception routes
- Autonomous deployment for strategic positioning
- Multi-vehicle coordination without radio chatter
- CARLA 0.9.15: Safe simulation environment
- ROS2 Humble: Sensor processing and vehicle control
- Zenoh: Ultra-low-latency messaging (<50ms)
- VSS: Standardized vehicle data format
- WebSocket: Real-time dashboard streaming
Combining LiDAR-based lock-on, autonomous pursuit, and VSS-standardized coordinate sharing enables multiple police vehicles to coordinate seamlessly without manual communication overhead.
Our team is divided into three specialized groups:
Chase Police Team - Implementing police vehicles that lock-on to criminal vehicles and pursue them relentlessly until encirclement is achieved
- DongMin Jang, MinHyeok Choi
Encirclement Police Team - Implementing police vehicles that receive lock-on signals and dispatch to the criminal vehicle's location via optimal routes
- JaeHong Lim, JeongMin Jo
Infrastructure & Integration Team - Implementing workload orchestration using Ankaios, backbone network configuration, web interface development, and overall coordination
- Juhun Kang
Each team develops and tests their code locally using CARLA 0.9.15. Once implementation and debugging are complete, we integrate all nodes together for system-wide testing.
- Testing
- Unit testing per team
- Integration testing
- Latency measurement for each component
- Documentation
- README documentation
- User guide
- Code Reviews
- Cross-team code reviews during integration
- Continuous feedback and iteration
We communicate face-to-face in our workspace, enabling rapid problem-solving and real-time collaboration. Additionally, we use messaging platforms for asynchronous updates and documentation sharing.
Approach 1: Team leader proposes ideas → Team members provide feedback → Implementation and testing
Approach 2: Team members propose ideas → Team leader and other members review → Implementation and testing
We maintain a collaborative decision-making process where all voices are heard, and decisions are validated through actual implementation and testing results.