Do Drivers Accurately Estimate Time-Required-to-Turn in Left-Turn Scenarios?

Abstract

While making left turns across traffic, drivers must judge whether they have enough time to complete the maneuver safely. This requires drivers to estimate time-to-collision of oncoming vehicles, and the time required to complete the left turn (TRT). Although prior studies measured the accuracy of time-to-collision judgments, less is known about judgments of TRT. Two experiments measured (1) drivers’ judgments of the last safe moment to turn in front of motorcycles and trucks travelling at 30 and 40 mph, and (2) drivers’ TRT judgments. Results indicated that drivers underestimated TRT, and a post-study questionnaire indicated that most participants believed their estimated TRTs were shorter than their actual left turns. Perceived safety margins (i.e., the judged last safe moment to turn minus estimated TRT) were larger for trucks than for motorcycles, and larger for slower vehicles compared to faster vehicles. Results have implications for the design of left-turn collision warning systems.

While making left turns across oncoming traffic, drivers must judge whether they have enough time to complete the maneuver safely. This requires drivers to estimate the time required for oncoming vehicles to reach them (time-to-collision or TTC) and the time required to complete the turn (TRT; Gray, 2004; Levulis, 2018). Although prior studies measured the accuracy of TTC judgments of oncoming vehicles, less is known about TRT judgments.

In addition, the size-arrival effect indicates that people perceive smaller vehicles as arriving later than larger vehicles (DeLucia, 1991) even during left-turn judgments (Levulis, 2018), which may increase the risk of collisions for left turns in front of small vehicles (e.g., motorcycles). Two experiments measured (1) drivers’ judgments of the last safe moment to turn in front of motorcycles and trucks travelling at 30 and 40 mph, and (2) drivers’ TRT judgments.

In Experiment 1, 24 licensed drivers used a STISIM Drive™ simulator to complete two tasks in counterbalanced blocks. In a turn-estimation block, participants verbally reported the start (gas pedal press) and end (cleared intersection) of imagined turns, repeated eight times. Then they actually turned the vehicle with the driving simulator controls eight times, followed by eight more imaginary turn estimations (see Hassan et al., 2007). Participants’ actual turn times were compared to their estimated turn times. In the judgment block, participants indicated the last safe moment to turn in front of an oncoming vehicle (motorcycle or truck) traveling at 30 mph or 40 mph. The estimated turn time was subtracted from the judged last safe moment to turn to determine the perceived safety margin.

Results indicated that drivers significantly underestimated TRT compared to their actual turn times. Perceived safety margins were significantly larger for trucks compared to motorcycles, which is consistent with the size-arrival effect (DeLucia, 1991). Perceived safety margins were also significantly larger for vehicles travelling at 30 mph compared to 40 mph. These findings may have been due to our instructions for the turn-estimation block, that the turn started at the first moment the gas pedal was pressed; but participants frequently began their movement by releasing the brake.

In Experiment 2, 24 different licensed drivers completed the same turn-estimation block from Experiment 1, except that they were instructed that the turn started when the vehicle began to move. Nevertheless, participants significantly underestimated their turn times. In both experiments, most participants reported in a post-study questionnaire that their estimated turn times were shorter than their actual turn times. A possible reason for turn-time underestimations is that participants did not incorporate the time it took for the vehicle to accelerate.

Overall, results suggest that people underestimate TRT, which means they perceive they have more time to turn than they actual do. This increases the risk of collision. Further, people have shorter perceived safety margins in front of relatively smaller and faster vehicles. The implication is that driver assistance systems should incorporate oncoming vehicle size and speed to help drivers decide when to begin turning left. However, it is important to consider limits in the generalizability of our results due to the use of a driving simulator, and passive judgments rather than active driving amidst oncoming (simulated) traffic.