Project Overview
A production-grade vehicle detection system deployed at multiple field locations to protect bridge infrastructure from over-height vehicle collisions. The system integrates multiple sensor types into a unified detection and alerting platform with real-time response capabilities.
System Architecture
Containerized multi-process application deployed on industrial hardware at roadside detection sites. The architecture emphasizes fault tolerance, automatic recovery, and minimal operator intervention.
Height Detection"] VCA["Video Analytics Camera
VCA Metadata via MQTT"] PTZ["PTZ Camera
RCP+ Protocol"] end subgraph PROCESSING["Processing Layer - Docker Container"] IPC["MQTT Message Bus"] PROC["Event Processing
Correlation · Deduplication"] DB[(PostgreSQL)] WEB["Flask Web Interface"] WATCHDOG["Watchdog Supervisor"] end subgraph OUTPUT["Output Layer"] RELAY["Relay Controllers
TCP/Telnet"] VMS["VMS Warning Signs"] FEED["Data Feeds
SFTP · NiFi"] DVS["Video Streaming Service
RTSP Relay"] UI["End-User Dashboard"] RTSP_OUT["Video Recording
RTSP Capture"] RELAY --> VMS FEED --> UI DVS --> UI RTSP_OUT --> UI end PTZ -->|"RTSP"| DVS VCA -->|"RTSP"| DVS LIDAR -->|"XML/TCP"| IPC VCA -->|"MQTT"| IPC PTZ -->|"RCP+ · RTSP"| PROC IPC --> PROC PROC --> DB PROC --> WEB WATCHDOG -.->|"monitors"| PROC PROC --> RELAY PROC --> FEED PROC --> RTSP_OUT
System Capabilities
Sensor Integration
Real-time data acquisition from multiple sensor types including LiDAR height detection and video analytics cameras.
Event Processing
Detection event handling with deduplication, correlation, and database persistence for audit trails.
Automated Response
Trigger-based activation of warning systems and video recording upon detection events.
System Reliability
Multi-tier health monitoring with automatic recovery mechanisms for continuous operation.
Technology Stack
Backend
Frontend
Infrastructure
Protocols
Hardware Integration & Protocols
Direct integration with industrial sensors and cameras required implementing multiple vendor-specific and standard protocols:
LiDAR Integration
Real-time height measurement data acquisition via TCP socket connections. XML-based event parsing with configurable detection thresholds and zone definitions.
Bosch RCP+ Protocol
Implementation of Bosch's proprietary Remote Control Protocol for PTZ camera control. Includes preset positioning, continuous movement commands, and camera status monitoring.
Video Analytics (VCA)
MQTT-based subscription to camera metadata streams. Real-time parsing of detection events with object classification and tracking data extraction.
RTSP Stream Management
Video stream capture and recording triggered by detection events. FFmpeg-based processing with automatic cleanup and FTP upload of evidence footage.
Relay Controller Interface
TCP/Telnet communication with industrial relay boards for VMS warning sign activation. Includes connection pooling and automatic reconnection logic.
SNMP Monitoring
Network device health monitoring via SNMP queries. Status polling for cameras and network infrastructure with alerting on failures.
Engineering Scope
The project required addressing challenges across multiple engineering domains:
- Multi-process coordination and inter-process communication
- Real-time stream processing with fault tolerance
- Event correlation across heterogeneous data sources
- Reliable operation in field environments with network instability
- Concurrent file access and cross-process logging
- Secure remote administration for distributed deployments
- Automated deployment and configuration management
- Integration with external traffic management systems
Deployment & Operations
Automated Installation
Custom installer with dependency management and system configuration for field deployment.
Configuration Management
Web-based settings interface with validation and secure credential storage.
Remote Updates
SSH-based deployment automation with rollback capability.
External Integration
Data feeds to centralized traffic management systems via standard formats.
Project Outcomes
Successfully deployed and operational across multiple field sites with continuous uptime requirements met.