Investigating OpenPilot as a Research Tool

Abstract

This report investigates the feasibility of using OpenPilot, an open-source driver assistance system, and a Comma 3, the hardware that runs this software, as a research tool for data collection and analysis at the Virginia Tech Transportation Institute (VTTI). The study explores OpenPilot’s data logging capabilities, camera views, and driver monitoring assessments while identifying potential applications and challenges for research integration. OpenPilot, when paired with Comma 3/3X hardware, offers real-time logging of vehicle kinematics, GPS data, and video feeds, making it a potential alternative to traditional data acquisition systems. This study successfully developed methods to extract and integrate OpenPilot data into VTTI’s research pipeline, enabling seamless analysis of driver behavior and advanced driver assistance system performance. A pilot evaluation demonstrated OpenPilot’s ability to collect high-quality data that is suitable for analyzing Level 2 systems usage, driver-initiated disengagements, and driver monitoring system (DMS) effectiveness. The study also conducted a structured assessment of OpenPilot’s DMS, revealing that while it effectively detects gross head movements, it struggles to identify subtle eye glances, limiting its reliability for distraction research. Additionally, concerns regarding data privacy and storage limitations on newer Comma devices present potential barriers to large-scale deployment. Despite these challenges, OpenPilot shows promise as a research tool with further development. Ongoing work at VTTI aims to address data privacy and storage issues while leveraging OpenPilot for the study of an intelligent speed assist system. With continued refinement, OpenPilot and Comma 3/3X devices could become valuable assets for cost-efficient driving data collection and advanced driver assistance system research.