RESUMO
Takeover is a critical factor in the safety of autonomous driving. Takeover refers to the action of a human driver assuming control of an autonomous vehicle from its automated driving system. This can occur when the vehicle encounters a situation it cannot handle, when the system requests the driver to take control, or when the driver chooses to intervene for safety or other reasons. This study explored how traditional steering-wheel driving habits affect takeover performance in joystick-controlled autonomous vehicles. We conducted an experiment using a joystick-controlled Dongfeng Sharing-VAN autonomous vehicle in a low-speed campus environment. The participants were divided into three groups based on their driving experience: the individuals who have no licence and no experience (NN Group), the drivers who have licence but not experienced (HN Group), and the drivers who have licence and have been experienced (HH Group), representing varying levels of driving habits. The experiment focused on two takeover tasks: passive takeover and active takeover. We evaluated takeover performance using takeover time and takeover quality as key metrics. The results from the passive takeover task indicated that traditional driving habits had a significant negative impact on takeover performance. The HH Group took 2.65 s longer to complete the task compared to the NN Group, while the HN Group took 3.78 s longer. When we analyzed takeover time in stages, the initial stage showed the most significant difference in takeover time among the three groups. In the active takeover task, driving habits did not significantly affect takeover braking in front of obstacles in a low-speed driving environment. These findings suggest that conventional driving habits can hinder passive takeover in joystick-controlled autonomous vehicles. This insight can be valuable for developing training programs and guidelines for drivers transitioning from conventional to autonomous driving.
RESUMO
OBJECTIVE: Driving comfort is crucial for tunnel safety because tunnel sections on freeways often introduce significant environmental changes that can compromise comfort and increase the risk of traffic accidents. This study aimed to quantitatively evaluate the driving comfort in tunnel sections and its implications for safety management. METHODS: Four indicators were used to assess the driving comfort: heart rate growth rate (Hrgr), skin conductance response (SCR), speed, and acceleration. The CRITIC weighting method was employed to calculate a quantitative driving comfort score, and the presence and severity of discomfort were used to evaluate the safety of each tunnel area. In addition, the evaluation was based on a naturalistic test consisting of Hrgr, SCR, speed, and acceleration data. A total of 32 participants were recruited based on a web-based questionnaire screening process, after which they were tested while driving through 30 tunnel sections on the roadway. These 30 tunnels included 14 short (< 500 m), 12 medium (500-1,000 m), and 4 long (1,000-3,000 m) tunnels. RESULTS: The results revealed that the four selected indicators exhibited minimal multicollinearity and effectively captured the driving comfort. Among the indicators, SCR had the most significant contribution to the driving comfort score. Most drivers did not experience substantial discomfort while driving through tunnels. The area where drivers were most susceptible to discomfort was the middle zones of tunnels. However, drivers were more likely to experience strong discomfort in the outside exit, entrance, and middle zones of short, medium, and long tunnels, respectively. CONCLUSIONS: This study provides a comprehensive set of safety evaluation methods for tunnel sections on freeways, with a focus on quantifying the driving comfort. The findings provide theoretical support for freeway management personnel in implementing personalized controls in different tunnel areas with the aim of enhancing tunnel safety and mitigating the occurrence of traffic accidents.