Data Integration Setup for Instrumented Vehicle - Summer Associate

Context

An instrumented vehicle was needed to collect naturalistic driving data for analyzing real-world driver behavior patterns. The Transportation Engineering Laboratory at IIT Kanpur required a cost-effective, scalable solution under Prof. Pranamesh Chakraborty's guidance.

Problem

• High production cost (INR 300,000) limited number of vehicles that could be instrumented

• Complex sensor integration required careful synchronization across multiple data streams

• Setup needed to be compatible with different vehicle platforms and architectures

• Lack of a standardized pipeline for timestamped multi-sensor data collection

Approach

Designed and deployed a modular sensor integration platform:

• Sensor integration — Evaluated specifications and integrated cameras, LiDAR, OBD, and GPS for synchronized data capture

• Data collection stack — Built Raspberry Pi-based system to collect, timestamp, and store sensor data in organized directory structures

• System optimization — Reduced hardware footprint and power consumption while maintaining data integrity

• Multi-vehicle compatibility — Designed setup to work across different vehicle models without major reconfiguration

Key Achievements

• Cost reduction — Lowered production cost from INR 300,000 to INR 15,000 (95% reduction)

• Scalability — Made setup compatible with multiple vehicle platforms

• Data integrity — Synchronized multi-sensor data streams with precise timestamps

• Reproducibility — Created a repeatable deployment process for rapid vehicle instrumentation

Learnings

• Hardware-software integration requires careful planning of power budgets and synchronization protocols

• Sensor fusion workflows benefit from standardized data formats and timestamp alignment

• Cost optimization often comes from architectural simplification rather than component selection alone

• Field deployments need robust error handling and remote monitoring capabilities

Impact

The reduced-cost platform enabled the lab to instrument multiple vehicles simultaneously, accelerating data collection for driving behavior research and naturalistic driving studies.