Autonomous detection of distracted driving by cell phone

Driver distraction is a major factor in loss of life and property on our nation's highways and the broader transportation systems. The role of wireless devices as a source of distraction is well established, significant, and growing. The potential of enabling such devices with the intelligence to detect distracted cognitive states of vehicle operators is of significant interest. This paper describes a concept that enables cell phones to autonomously detect distracted driving behaviors associated with texting. Unlike conventional methods, this detection paradigm measures how texting performance is affected by driving instead of how driving performance is affected by texting, which is well documented. This new approach can be extended to other device inputs such as speech and is compatible with a spectrum of countermeasure actions to mitigate the source of distraction. A cell phone was programmed to log keystroke dynamics using a common operating system. This platform was used to characterize the texting dynamics of six subjects. Study participants were observed texting alone and, during a separate session, while texting and operating a driving simulator. This study yielded reliable distracted driving signatures that are independent of the explicit communications language and text content. This paper discusses the results in light of acknowledged distracted driving challenges, the potential of using the method for autonomous mitigation, and the speed of classification.