Lytx’s MV+AI technology explained – how the DriveCam detects risks

Lytx

how does ai work

The machine vision and artificial intelligence technology behind Lytx “triggers”

The innovative DriveCam® helps improve fleet safety by watching for risky driving behaviors on the road and in the vehicle. The device’s advanced machine vision (MV) and artificial intelligence (AI) technology analyzes driver behavior and nearby vehicles to determine how a driver is performing relative to their surroundings. Specific driver behaviors such as following too closely, failing to obey a stop sign, or texting while driving, trigger the DriveCam to flag and “trigger” the event.

What are the benefits of MV+AI technology?

Lytx’s MV+AI technology helps fleets by providing an expanded view of risk. The insights from MV+AI can reveal if a driver needs to work on reducing a certain behavior or if certain risky behaviors are appearing among all drivers. If this is the case, an organization-wide initiative to improve safety may need to be implemented.

 

How the MV+AI technology learns

Lytx’s MV+AI technology uses information from several sensors at once to identify risky situations and driving behaviors. Machine vision can see and recognize objects and behavior by analyzing images and video data. AI interprets and learns from those images and video data to determine the likelihood that a particular event or behavior occurred. The combination of information from video and other sensors in and around the vehicle, called video telematics, provides information to help the artificial intelligence learn—similar to the way our brains rely on information from each or our senses to understand what is happening around us.

Not just any data can be used to train the AI to identify and categorize risky driving behaviors. The data must come from a large database. It also must be validated. The most accurate validation comes from human reviewers. Here’s why:

Human experts help surface important nuances that help the system learn faster and interpret more complex scenarios. For example, AI technology might detect a lane departure and trigger a video event clip. We know that lane departure is highly correlated with distracted driving such as texting.  However, a human reviewer might notice that this swerve was due to a construction area forcing all traffic to make a sudden lane change. This type of nuanced data is important in helping to refine, learn, anticipate what might happen next and become more sophisticated along the way.  Lytx events have been reviewed by humans for over 10 years. 

 

How do MV+AI triggers work?

Machine Vision recognizes an object through the DriveCam event recorder’s lens, and AI uses that information, along with other data like speed, GPS to determine if that combination of information is risky. You can think of machine vision as the eyes, and AI as the brain.

For example, when a vehicle approaches a stop sign, machine-vision technology scans the sign and recognizes it. The system also uses speed data to detect whether the driver came to a complete stop. If the accelerometer detects that the vehicle didn't come to a complete stop the DriveCam system triggers a video clip.

 

Here’s a list what the DriveCam can currently detect on the road using Lytx’s MV+AI :

Rolling stop:

Following distance:

Critical distance:

Lane departure:

 

Driver-facing lens detects distraction

Lytx’s MV + AI technology is trained to recognize objects and actions that indicate distracted driving behavior such as using a cell phone, eating and drinking, smoking, or failing to wear a seatbelt. The driver-facing camera captures and analyzes images that allow the machine vision technology to recognize an object such as a cigarette or a cell phone. The AI is trained to categorize the cell phone, cigarette, food, drink, or seat belt use as risky.

Even if the object in question is hidden from view, the AI can detect behaviors such as looking down repeatedly to determine reliably that a driver is texting, for example. Importantly, risky distracted driving behaviors tend to happen together. Almost 1 in 4 drivers who are engaging in one risky behavior are actually doing more than one at the same time.

 

Identifying and coaching risky behavior is one of the most important steps in improving overall fleet safety. Lytx’s fleet management solutions use machine vision and artificial intelligence technology that helps uncover previously undetected risky driving behaviors so that fleet managers can coach their drivers to improve. See how your team can benefit from this revolutionary technology, get a free demo.

 

Previous
Distracted Driving Statistics and Tips
Distracted Driving Statistics and Tips

Take a look at some chilling stats on distracted driving, and get some actionable tips to help you reduce d...

Next
4 Dash Cam Videos that Show the Dangers of Distracted Driving
4 Dash Cam Videos that Show the Dangers of Distracted Driving

Watch dash cam videos showing some of the most common examples of distracted driving and the dangers that c...

See all resources on improving driver performance

View all