Author: Sasha Tanner
Law enforcement uses the Standardized Field Sobriety Test (SFST) on suspected impaired drivers to determine if they are under the influence of alcohol and to potentially use the evidence in court. What if this process were accompanied by a device that enhances the SFST process? Roshan and Park (2025) present the automated impairment detection system, a device that captures the suspect’s movements to measure their level of intoxication.
Source: iStock
Methods
First, the device was programmed to analyze the accuracy for the One Leg Stand (OLS) and the Walk and Turn (WAT). It was designed to detect alcohol and cannabis impairment with a camera, eye testing screen, biometric measurement sensors, and a body movement imaging system.
Participants displayed the following characteristics:
- Sex: 22 males and 12 females
- Sobriety: 17 sober, 14 alcohol-impaired, and 3 cannabis impaired
- Average Age: 29.98 years
Then, participants engaged in the SFST as the device recorded them. They performed the OLS and WAT tests, in particular.
- OLS: Participants were required to lift one leg approximately 6 inches away from the ground and maintain that stance for 30 seconds with arms at the sides, counting from 1,000 to 1,030. The same routine goes for the opposite leg.
- WAT: Participants were required to walk in a straight line for a few yards in a heel-to-toe fashion before turning around and repeating the process without stumbling.
The device captured 14 key points in their bodies that targeted joint movements. More specifically, the researchers measured the following key points with mathematical formulas:
- Scale Value: This accounts for the various distances from the camera by measuring the Euclidean distance between 2 key points, such as the shoulder center and hip center.
- Euclidean Distance Between Key Points: Across time, these measures will fluctuate if the participant cannot maintain a balanced position.
- Balance Ratio: This formula measures the neck center and hip center to determine vertical alignment over a distance.
Results
On each test, participants’ key points fluctuated more during impaired SFSTs than for sober ones; these data were collected to analyze which joints signaled impairment.
OLS Test
- Right Hand, Right Foot: Results were less relevant because usually, this is a person’s dominant, more stable side.
- Left Hand, Left Foot: These results were more relevant because if the participant was impaired, they would have a difficult time balancing.
WAT Test: There were 3 peak values that the researchers analyzed to capture joint movement and oscillation levels that signaled impairment.
- Right Hand, Right Leg Joints: Associated with the walking portion of the WAT test.
- Neck, Left Hand Joints: Associated with the walk-back portion of the WAT test.
- Neck Center and Hip Center: These key points were recorded during the process through the balance ratio formula.
In short, the right hand and foot are critical to examine impairment during the OLS test if that is their dominant side. The WAT test is more complex, where the right-hand side is more relevant walking forward while the neck, left-hand side, and hips signal balance during the walk-back portion. Also, results show that the right-hand side for the OLS test is more consistent with indicating impairment than the WAT test.
Recommendations
- Law enforcement officers could use this automated impairment detection system device while they conduct the SFST on a suspected impaired driver. In this way, they have quantitative data to supplement and confirm their case during a court hearing.
- This device could be programmed to look for key points that signal cannabis or other forms of drug impairment to help law enforcement conduct the SFST.
- This device could be programmed into law enforcement body cameras as they delegate the SFST on a suspected impaired driver.
Technology is supposed to assist people in carrying out complex tasks in an efficient way without replacing their expertise. If you are interested in reading more about the study, check out the source below!
Source: Roshan, S. M., and E. J. Park. (2025). Automated Vision-Based Detection of Impairment Through Divided Attention Psychophysical Tests. IEEE Sensors Letters, 9(9), 1-4, Article no. 6009404, doi: 10.1109/LSENS.2025.3594392
