Drivers often engage in secondary in-vehicle activity that is not related to vehicle control because they believe they can do so safely. Often, it may be to relieve the monotony of driving. Interest is growing to understand and measure a driver’s workload, and design vehicle functionality to accommodate a driver’s perceived, rather than actual, workload. An accurate and real-time variant measure of driver workload that is personalized to an individual driver could be useful in the design of vehicle functionality that can be invoked and brought to the foreground when necessary, or placed in the background when not necessary. In autonomous vehicles where a driver is present as part of the HMI (human-machine interface), this structure could be helpful to better understand the transition from automated to manual driving mode, and vice versa. In this study, the measurement of perceived workload, and its inherent ‘personalized’ connotation was investigated. In earlier studies, we had reported on the development, use and evolution of a Transportable Instrument Package to record vehicle, environmental, and physiological data, and subsequent selection and classification of parameters and measures that could be relevant to further computational investigations. In this paper, we describe further enhancements to the data collection system, and on-road data collection and algorithms to determine perceived workload. The basic three major modules were retained. There was also a GPS recording device, and two video cameras, one to record the forward path, and the second to record the driver’s face and direction of gaze. An expert human driver served as a ride-along passenger to provide an estimate of driver workload, and record the perceived workload as reported by the subject driver. The mobile app module was developed to record traffic events (turns, lane changes, traffic signs). In this paper, we explain how the mobile app was designed and show examples of the types of data that have been collected. We also present preliminary results on the performance of the driver workload estimation algorithms developed.