The Next AI Benefit: Better Driver Scorecards

Driver scorecards have been used for over a decade for a variety of purposes ranging from measuring idling, number of trips, ELD compliance and time on site to identifying drivers who present a significant safety risk, but they have been an imperfect solution for safety monitoring. Drivers dinged for factors such as speeding, aggressive acceleration and harsh braking typically challenge the point deductions by claiming they were merging onto a busy on-ramp, braking on a steep mountain road, or otherwise driving in a normal manner. Fleets, in turn, have had no way to verify or disprove those claims.

These sorts of “he said, she said” confrontations have had two serious repercussions.

First, they have undermined trust in scorecards as a measure of driver performance. That has prompted many fleets to simply disregard the scoring reports, ignoring their value as a tool for pinpointing high-risk drivers in need of coaching.

Second, these disagreements have caused unnecessary conflict with drivers who are penalized for events that have legitimate explanations. This raises the threat of driver resignations at a time when the industry is already short an estimated 80,000 drivers and facing a deficit of more than 160,000 by 2030.

Benchmarking has helped upgrade scorecarding systems by comparing a fleet’s performance to operations of similar size, industry, geography and vehicle makeup, but it doesn’t solve the fundamental problem of whether a driver deserves to be docked for a given event flagged by the telematics system. It can provide useful insights into whether a fleet’s drivers are performing better or worse than average, but it fails to improve the accuracy of the scoring system on which a driver’s future as well as a fleet’s safety record may rise or fall.

Getting It Right

At the heart of the problem with safety-related scoring errors is the inability of standard telematics technology to distinguish between risky driver behavior and actions taken in response to anomalies like road, weather and traffic conditions. Telematics devices use G-sensor and GPS data to gather information on issues such as vehicle speed, acceleration rate and hard braking, but they are literally blind to variables outside the cab.

Today’s most advanced video telematics cameras are removing these blinders not only by adding video documentation but also by applying a combination of machine learning, machine vision and artificial intelligence to analyze the footage.

This contextual analysis makes it possible to determine whether an event captured by a dashcam needs further investigation or intervention to ensure fleet safety. It differentiates between a pothole and a collision; a sharp curve and a collision avoidance maneuver; an expected acceleration and an unsafe speed increase; braking required for downhill driving and harsh braking suggesting unsafe vehicle operation; and more.

One of the first solutions capable of making these distinctions is SmartWitness AIDE (Artificial Intelligence Driving Events), cloud-based AI-powered software that has been included with every SmartWitness dashcam for the last year. AIDE analyzes each event in near-real time and automatically filters out those that can be explained by local terrain or other real-world driving challenges, reducing false alerts sent to fleet managers for review as much as tenfold. Over time, the software’s machine learning also learns an individual driver’s normal driving patterns, further refining the ability to evaluate driver performance.  

While the results of these advanced solutions are not yet integrated into driver scorecards, fleet managers are already reaping the benefits. AI’s ability to reduce false alerts is resulting in:

• Less time wasted in unnecessary alert review and driver coaching sessions
• Increased fleet manager productivity
• Less conflict with drivers falsely accused of risky behavior behind the wheel
• More time focused on drivers who present a true safety risk

Improving Scorecard Effectiveness

With AI-powered intelligent video event analysis in place, TSPs will have the ability to develop next-generation scorecards that overcome the limitations of today’s scoring systems in accurately reporting safety infractions, including distracted driving.

If a driver has to slam on the brakes in order to avoid an accident, or if the telematics device mistakes a rough road for a close encounter with another vehicle, the scorecards of the future will be able to recognize the difference and avoid assigning unfair demerits that both weaken scorecard value and chip away at driver morale. This also promises to increase drivers’ willingness to listen to constructive feedback.

Equally important, newer scorecards will be able to incorporate video documentation of flagged events that will provide irrefutable evidence of safety violations while also serving as a useful aid for coaching. Just as professional athletes use video clips of themselves as training tools for improving their performance, drivers will be able to view the offenses that triggered point deductions and use the video as a guide for taking corrective action.

This broad range of new visibility features – all enabled by core video telematics technology combined with recent AI-related advances – will bolster the efficacy of scorecards in helping fleets optimize safety and strengthen working relationships with drivers.

Given the urgent need to address the driver retention crisis as well as protect fleets from accidents that can lead to staggering legal costs, every step toward better scorecarding is a step toward a better business environment for fleets everywhere.

About the Author: Michael Bloom is Vice President of Product and Marketing for SmartWitness, a global provider of video telematics solutions that help fleets optimize operations, improve driving behavior and mitigate risk. SmartWitness was recently acquired by Sensata Technologies, a global industrial technology company striving to create a cleaner, more efficient, electrified, and connected world.