Abstract
The connected vehicle systems (CVS) aim to provide drivers with information in a timely and reliable way to improve transportation safety. With the emerging wireless communication technologies, the vehicles will be equipped with the ability to communicate with each other about the surrounding traffic situations by exchanging vehicle status and motion data via Dedicated Short-Range Communications (DSRC) network (Kenney, 2011).
With the assistance of the cooperative collision warnings, the impact of designed warning parameters on driver performance is increasingly important. Existing empirical studies have studied the warning timing and warning reliability in determining the effectiveness of the collision warning systems in advanced driver assistance systems (ADAS). In terms of warning timing, the studies in reached consistent conclusions that early warnings induced more timely braking and longer braking process, resulted in higher trust of the warning systems, and reduced collision rates (for example, Abe & Richardson, 2006a; Lee, McGehee, Brown, & Reyes, 2002; Yan, Xue, Ma, & Xu, 2014; Yan, Zhang, & Ma, 2015; Wan, Wu, and Zhang, 2016). In terms of the warning reliability, research has shown that warnings with a higher reliability increased driver’s trust of the warning systems, led to higher frequency in warning responses, and reduced crash rates (for example, Abe, Itoh, & Yamamura, 2009; Bliss & Acton, 2003; Maltz & Shinar, 2007; Sullivan, Tsimhoni, & Bogard, 2008). However, the interaction effects of warning lead time and warning reliability on driver performance was not examined especially under the connected vehicle settings.
The current research investigated the interaction effects of warning lead time (2.5s vs. 4.5s), warning reliability (73% vs. 89%), and speech warning style (command vs. notification) on driver performance and subjective evaluation of warnings in CVS. A driving simulator study with thirty-two participants was conducted to simulate a connected vehicle environment with missing warnings due to the failures in the data transmission within the communication network of the CVS.
The results showed command warnings led to a smaller collision rate compared to notification warnings with the warning lead time of 2.5s, whereas notification warnings resulted in a smaller collision rate compared to command warning with the warning lead time of 4.5s. These results suggested notification warnings should be selected when warning lead time is longer and warning reliability is higher, which resulted in higher safety benefits and higher subjective ratings. Command warnings could be selected when warning lead time is shorter since they led to more safety benefits.
However, such selection has to be made with caution since command warnings may limit drivers’ response type and were perceived as less helpful than notification warnings.