The Impact of Electrification and Partial Automation on Driver Speeding Behavior

Abstract

As electric vehicles (EVs) and partial automation systems become increasingly prevalent, their impact on everyday driving behavior remains underexplored. This study utilizes real-world naturalistic data to examine how vehicle type, an electric versus an internal combustion engine (ICE), and the use of partial automation are associated with speeding behavior. Data were collected from 24 drivers over the course of a month each, comparing Tesla Model 3s with Autopilot (EV) and Cadillac CT6s with Super Cruise (ICE), covering about 38,000 miles of driving. Results indicate that EV drivers tended to speed for shorter durations on arterial roads but exhibited higher speeding magnitudes on residential and controlled access roads after their first week of driving. Notably, driving with partial automation, regardless of powertrain, was associated with significantly longer speeding durations and slightly greater speeding magnitudes compared to manual driving. These findings suggest that both electrification and automation contribute to evolving driver behaviors, changing speeding behavior in specific driving contexts. As drivers adapt to new vehicle technologies, understanding how these systems shape behavior is important. Insights from this study may inform the design of future in-vehicle systems and guide driver education strategies to promote safe driving practices in an evolving transportation landscape.

Introduction

Speeding has been shown to be a prevalent risky behavior that not only increases the likelihood of a crash but also the severity of the injury if a crash were to occur (Aarts & van Schagen, 2006; Elvik, 2019; Elvik et al., 2019). Since March 2020, documented driving patterns and behaviors indicate a general increase in average traveled speeds and that extreme speeds (20 MPH or higher than the posted speed) have become more common (NHTSA, 2021). In 2023, although not causally related to every outcome, speed was a factor in 11,775 fatalities (29% of US traffic fatalities) and 12% of traffic injuries (NHTSA, 2025).

Another relevant trend is the increase in electric vehicles (EVs) in the US, which, as of 2024 were 21% of total light-duty vehicle sales (Abboud, 2024). The introduction of EVs, coupled with the increased availability of driving automation systems, may have a broad influence on driver behaviors, including speeding, driving kinematics, and subsequent risk outcomes. To date, the discussion on the impact of EV adoption is focused primarily on greenhouse gas emissions and crashworthiness, while other aspects of driving EVs - including construction, performance characteristics, and operation - have received less consideration despite their potential to impact driver behavior.

Paralleling the rise of electrification, the automotive industry has witnessed a surge in the development and implementation of driving automation systems. Consumer vehicles currently on the market can be equipped with partial automation systems (SAE Level 2) that simultaneously control the longitudinal (e.g., adaptive cruise control) and lateral (e.g., lane centering) vehicle kinematics on a sustained basis (SAE International, 2021). Prior studies have found associations between partial automation use and changes in driver behavior and driving kinematics, including speeding prevalence (Gershon et al., 2021, 2024; Haus et al., 2022). These associations are particularly relevant as Advanced Driver-Assistance Systems (ADAS) technologies are becoming ubiquitous in the US, with adaptive cruise control (ACC) available in 92% of new car models and 50% of models offering Level 2 ADAS (Consumer Reports, 2021).

Together, the concurrent increase in the availability of EVs and partial automation is raising questions about how these technologies jointly shape speeding behaviors. This study evaluated the associations between vehicle type (ICE or EV), partial automation use, and speeding behavior across different driving environments and within the first month of interaction, providing details on how drivers may adapt to driving EVs. Using naturalistic driving data, we analyzed how speeding duration and magnitude varied between individuals driving a representative EV and an ICE vehicle, and how these behaviors changed over time. By identifying differences in speed control habits and adaptation processes, our findings provide insights into how emerging vehicle technologies may shape driver behavior. This understanding may inform the development of future driving systems, effective driver education strategies, and policies to enhance safety and promote desired behavioral change.

Methods

Data were drawn from the ongoing MIT-AVT naturalistic data collection effort (Fridman et al., 2019). The analytical dataset included a total of 2,441 trips, covering 17,674 miles for EV and 20,258 miles for ICE vehicles. Data were collected from 24 drivers, 12 EV drivers (mean age = 39 years, SD = 12, 7 Female) who drove a Tesla Model 3 with Autopilot, and 12 ICE drivers (mean age = 43 years SD = 13, 3 Female) who drove a Cadillac CT6 with Super Cruise. Each study vehicle was instrumented with RIDER (Real-time Intelligent Driving Environment Recording), a custom data acquisition system that collected data from four camera views, GPS, and the vehicle’s CAN-Bus. The dataset included vehicle speed (mph) captured directly from the CAN-Bus; road type and speed limit were derived from the GPS data.

The different vehicle accel/decel values were binned to the following:

• Elevated Driving Kinematic: g-force ≤−0.45g

• Strong decel.: g-force (−0.45, −0.3]

• Weak decel.: g-force (−0.3, −0.15]

• Drag torque: g-force (−0.15, −0.05]

• Cruising: g-force (−0.05, 0.05]

• Accel.: g-force (0.05, 0.4]

• Strong accel.: g-force >0.4g

Road types were binned into three categories:

• Controlled access: includes interstates and divided highways

• Arterial: includes primary, secondary, and tertiary arterial roads, major and minor throughways, which are not divided highways

• Residential: includes minor roads in residential areas.

Lastly, speeding behavior was defined as driving above the speed limit for at least three consecutive seconds. The analysis assessed the speeding duration and magnitude (speed above the speed limit), as a function of vehicle type (ICE and EV), automation use (manual or partial-automation), and changes over time (study weeks). All data were analyzed at a 10 Hz sampling rate.

Results

A generalized linear mixed-effects model with a gamma distribution and driver-specific random intercept was used to assess differences in driving exposure across roads and vehicle types. In general, the overall driving exposure of EV and ICE drivers were relatively similar with EV drivers having slightly longer trips on arterial and controlled access roads (arterial Est.Mean: 5.3 miles/trip 95% CI[4.4–6.4]; controlled access Est.Mean: 25.8 miles/trip, 95% CI [21.2–31.3]) compared to ICE drivers (arterial Est.Mean: 4.1 miles/trip, 95% CI[3.4–4.9] controlled access Est.Mean: 18.6 miles/trip, 95% CI[15.3–22.7]). To characterize the differences in driving kinematics between EV and ICE vehicles, we assessed the prevalence of different acceleration categories, with and without use of automation, and across road and vehicle types using a Linear Mixed Effects model with full interaction terms and a driver-specific random intercept. In general, EV driving was characterized by a higher prevalence of the extreme accel./decel. categories. In manual driving, EV drivers had a significantly higher prevalence of strong accelerations (>0.4g) and strong decelerations (≤–0.3g) compared to ICE drivers. Cruising was the most common acceleration type, occurring over half the time for both vehicles, especially when using automation. Cruising increased significantly for both EV and ICE when using partial automation compared to manual driving, with a greater difference for EVs (+0.17, p < .001, 95% CI [0.16–0.19]) than ICE vehicles (+0.14, p < .001, 95% CI [0.13–0.16]) (see Figure 1).

Figure 1.

Prevalence of each acceleration condition by vehicle, road type, and level of automation.

Generalized linear mixed-effects models with a lognormal distribution and a driver-specific random intercept were used to estimate the duration and magnitude of speeding. In general, driving on arterial roads was associated with 19% longer speeding duration among ICE drivers compared to EV drivers (Est.Mean ICE: 20.4 s, 95% CI [18.4–22.6], Est.Mean EV: 17.1 s, 95% CI [15.4–18.9]).

Analysis of speeding duration indicated that EV drivers sped for shorter durations than ICE drivers on arterial roads starting from the first week in the study, but there was no significant difference in speeding duration between ICE and EV drivers on arterial roads by the end of the study (see Figure 2). There was little difference between ICE and EV drivers in speeding duration on residential and controlled access roads.

Figure 2.

Estimated speeding duration by vehicle and road types over time in the study. *Indicates on p-value smaller than .05.

EV drivers had greater speeding magnitude on residential (EV: 5.43 mph, 95% CI [4.84–6.10] ICE: 3.40 mph, 95% CI [2.93–3.96]) and higher speeding magnitude on controlled access roads (EV: 3.96 mph, 95% CI [3.63–4.31], ICE: 3.49 mph, 95% CI [3.20–3.81]) compared to ICE drivers.

Figures 3 shows how differences in the magnitude of speeding evolved over the study weeks. After the first week of driving, EV drivers exhibited higher speeding magnitudes on residential and controlled access roads compared to ICE drivers. The difference in speeding magnitude between EV and ICE in residential driving increased throughout the study, with EV drivers averaging an estimated 1.39 times (p < .05) greater speeding magnitude in week 2, 1.64 (p < .01) greater magnitude in week 3, and 1.85 (p < .01) greater magnitude in week 4.

Figure 3.

Estimated speeding magnitude (amount over speed limit) by vehicle and road types over time in the study.

As Cadillac’s Super Cruise is a geofenced partial automation system that is available only on controlled access roads, the associations between partial automation use and speeding behavior were evaluated only on this road type. Figures 4 and 5 show speeding duration and magnitude on controlled access roads in manual and when driving with partial automation. Driving with partial automation was associated with almost 3 times longer speeding durations (partial automation/manual driving ratio: 2.9, 95% CI [2.7–3.1], p < .001) and had a slightly higher speeding magnitude (partial automation/manual driving ratio: 1.3, 95% CI [1.2–1.3], p < .001). However, no significant differences were found between EV and ICE vehicles.

Figure 4.

Estimated speeding duration by vehicle and road type.

Figure 5.

Estimated speeding magnitude (amount over speed limit) by vehicle and road type.

Discussion and Conclusions

The growing adoption of electric vehicles, coupled with the increased availability of partial automation, may bring significant changes to driver behavior. In this study, EV drivers exhibited a higher prevalence of strong acceleration (>0.4g) and strong deceleration (≤−0.3g) compared to ICE drivers. Cruising (i.e., minimal acceleration or deceleration) was the most frequent state for both vehicle types, especially when using automation. Notably, automation use was associated with more cruising time in EVs than in ICE vehicles.

Speeding behavior differed between a sample of EV and ICE drivers. Compared to ICE drivers, the EV drivers tended to speed for less extended periods on arterial roads and were more likely to exceed speed limits by greater margins on residential and controlled access roads. Differences in speeding duration between EV and ICE drivers on arterial roads were evident from the start of the study. However, differences in speeding magnitude on residential and controlled access roads emerged after the first week, reflecting a gradual shift in speed control among EV drivers. Additionally, the findings indicate that regardless of powertrain, the probability of speeding appears to increase with automation use (Haus et al., 2022), which may or may not be a result of the presence of the automation software. By identifying patterns of how increased speeding ties to both vehicle type and automation use, this work provides insights that can be used in the development of proactive strategies to enhance safety early in the technology adoption process. Future work may consider other vehicle models to assess the generalizability of these findings.

In conclusion, this study illustrates differences in driving behavior across powertrains and automation use that may be incorporated into future system design. While these technologies offer clear environmental and comfort benefits, they may also introduce shifts in driver behavior. As more drivers transition to EVs and gain access to partial automation, additional research may generate insights to support designs that further enhance safety. The results of this study have practical implications for the development of driver training programs, safety messaging, and vehicle system design.