Invented by Yeung; Fai, Krishnan; Abhiram, Chan; Brian K., Rivian IP Holdings, LLC

Vehicles today, especially electric vehicles (EVs), come with cameras that record nearly every moment on the road. But storage on these vehicles is not endless, and as drives stack up, important clips can be lost. This article will break down a clever way to save the best video moments using smart scoring and AI, making sure special memories and key events are kept safe while freeing up space for new adventures. Let’s get started by understanding why this matters, how it all works, and what makes this idea truly unique.

Background and Market Context

More and more cars, trucks, and other vehicles now come with cameras built in. These cameras don’t just help drivers park or stay safe; they record the entire journey. For many people, these videos are fun keepsakes, a way to remember family road trips, fun outings, or exciting off-road adventures. Sometimes, they even capture important events like accidents or near-misses, which can be crucial for insurance or safety reviews.

But there’s a problem. The storage that comes with these vehicles isn’t limitless. Video files, especially high-quality ones, take up a lot of space. As drivers continue to use their vehicles, the cameras keep recording new footage. Eventually, the storage fills up, and the system needs to decide what to delete to make room for new videos.

Up until recently, most systems used a simple method: delete the oldest files first. But this can cause problems. Imagine if an important moment—like the only video of an accident, or a once-in-a-lifetime scenic view—gets erased just because it was recorded a while ago. That’s not helpful for drivers who want to keep their best memories or need proof of a certain event.

The demand for smarter video storage is growing. Drivers want to keep their most important or interesting videos. Car makers want to offer better features and stand out from the competition. Insurance companies, law enforcement, and even families on vacation all benefit from keeping the right footage. With new technology like artificial intelligence (AI), it’s finally possible to do more than just delete old files blindly.

What’s needed is a way for the vehicle to know which video clips matter most, and which ones can be deleted without regret. This is where the new system comes in, making use of the vehicle’s own data, smart scoring, and AI to make smart choices about what stays and what goes.

Scientific Rationale and Prior Art

To understand how this new system is different, let’s look at how things have worked before and why those ways are not enough.

Earlier vehicle video systems, like dash cams, have mostly used simple rules. The most common rule is “first in, first out”—the oldest video gets deleted first. Some systems allow users to “lock” certain clips, usually by pushing a button after an event happens. More advanced systems use sensors to make decisions. For example, if a crash is detected, the video at that time is locked and cannot be deleted. But outside of these special cases, everything else is treated the same.

There have also been efforts to split videos into small pieces, or “fragments,” to make them easier to manage. Some systems will keep certain fragments if a sensor detects something big, like a sharp turn or sudden stop. But these methods still use simple rules and don’t look at the whole picture.

Recently, some companies have started using cloud storage. Videos can be uploaded to the cloud, so storage space is not a problem. But this costs more money, uses more data, and raises privacy concerns. Most drivers still rely on the storage built into their vehicle.

Artificial intelligence has changed many industries, but it’s only just starting to be used for vehicle video management. Some newer systems use AI to detect scenes, like city streets or country roads, and use that to tag videos. But few, if any, use AI to help decide which clips to keep and which to delete, especially using real-time data from the vehicle itself.

The main problem with the old ways is that they don’t use all the information available. They don’t look at things like how fast the car was going, where it was, whether something interesting or unusual happened, or what the driver might care about most. They also don’t adjust their choices based on how full the storage is, or change their behavior as space runs out.

The new system described in this patent application does all of that. It uses data from the car—like speed, location, sensor readings, and more—to give each video fragment a “priority score.” It uses a smart formula to weigh how important each clip is, and it can even adjust its decisions as storage fills up. By using AI, it can also recognize what’s happening in each video—like if a crash, sharp turn, or special event has happened.

This is a big step forward: instead of just deleting the oldest clips or making a few special exceptions, the system can make smart, flexible choices based on what’s really important. It saves the memories and moments that matter most, and does it all automatically.

Invention Description and Key Innovations

Now, let’s break down how this new system works, what makes it special, and how it actually helps drivers in real life. We’ll walk through the main ideas, step by step, in simple terms.

1. Assigning Priority Scores to Video Fragments

Every video the car records is split into smaller pieces, called fragments. Each fragment gets looked at by the system, which uses information from the car itself to decide how important that fragment might be.

For example, if the car’s sensors show that there was a sudden stop, sharp turn, or a big bump, the video fragment at that time might get a higher score. If the car was just cruising down a straight road with nothing special happening, that fragment might get a lower score. The system can also use information like the car’s speed, its location, what mode it was in (like off-road or city driving), and even features detected by AI in the video—like a beautiful mountain scene or a busy intersection.

All of this information is used to give each video fragment a “priority score.” The higher the score, the more important the system thinks that video fragment is.

2. Storing Scores as Metadata

Once a fragment has its score, that information is saved as part of the video file itself, as “metadata.” This means the score always stays with the video fragment and can be used later whenever the system needs to decide what to keep.

3. Calculating Retention Values Based on Storage Space

The system is always keeping track of how much storage space is left. When space starts to run low, it doesn’t just delete the oldest clips. Instead, it uses a special formula—called a “retention value”—that looks at both the priority score and the amount of free space left.

If there is plenty of space, the system is less aggressive about deleting clips. But as space gets tighter, the formula changes. Clips with lower scores are more likely to be deleted, while high-scoring fragments are protected for as long as possible.

The formula uses a “decay” function, so the closer the storage gets to being full, the stricter the system becomes about what gets deleted. This means the system is always adapting, making sure the most valuable moments are kept, no matter how close the storage is to the limit.

4. Using AI to Recognize and Protect Special Events

When something big happens—like a crash, a sudden jolt, or a unique scene—the system’s AI kicks in. It can spot these events by looking at sensor data (like G-force or jerk intensity) and by analyzing the video itself.

When these events are found, the system can gather all the fragments from that time, even from multiple cameras. It then protects these fragments from being deleted, making sure the whole event is saved. The system can even stitch together these fragments into a single “event video” for easy viewing later.

5. Choosing the Best Fragments from Multiple Cameras

Modern vehicles often have more than one camera—front, back, sides, and more. The system can look at all the fragments from the same time across these cameras. If more than one camera catches the same event, the system uses AI to decide which view is best. For example, if one camera has a clear view of a crash, while another just shows an empty road, the system keeps the important one.

This helps avoid storing duplicate or boring footage and makes sure that the most useful clips are kept.

6. Making Event Videos and Highlight Reels

Because the system knows which fragments are most important, it can do more than just protect them. It can automatically create videos of key events—like a highlight reel of a road trip, or a video showing just the moments when something exciting happened. This saves drivers from having to hunt through hours of footage to find the best parts.

7. User Preferences and Customization

Not everyone cares about the same things. Some drivers might want to keep all off-road adventures, while others care more about city driving. The system allows for user preferences to be set, so the scoring and selection matches what the driver wants. For example, a user can set preferences to favor certain types of scenes, or to keep longer clips from the same camera view.

8. Smart Deletion and Summarization

When space gets low, the system doesn’t just delete files randomly. It can “prune” out the parts that are less interesting, summarize long videos by keeping only the highlights, and clear out duplicates. This keeps storage use as low as possible while holding onto the best content.

9. Easy Integration and Use

The system is designed to work with the hardware already in modern vehicles. It uses existing sensors, cameras, and storage devices. The AI models can run on the car’s own computer, or on a server if more power is needed. All of this happens in the background, so drivers don’t have to do anything special.

10. Real-World Benefits

For drivers, this means peace of mind. Important memories—like a child’s first road trip, a beautiful sunset, or proof of an accident—are much less likely to be lost. It also means less time spent sorting through dull footage, and more space for new adventures.

Car makers can use this system to offer smarter, more helpful features. Insurance companies can trust that key footage is kept safe. And all of this is done without the need for expensive cloud storage or complex manual controls.

Conclusion

In a world where cars are always watching, not every moment is worth saving—but some are priceless. The new system described here uses the car’s own data, smart formulas, and artificial intelligence to make sure the best memories and most important events are never lost, even as storage fills up. By giving each video fragment a score, adjusting choices as space runs low, and using AI to spot key moments, this invention brings peace of mind to drivers and opens up new possibilities for how we remember and use our journeys.

This approach goes far beyond old “delete the oldest” methods, offering a smart, flexible, and user-friendly way to manage vehicle video storage. As cars continue to get smarter, systems like this one will become an essential part of the driving experience—making sure that the stories we want to remember are always there when we need them.

Click here https://ppubs.uspto.gov/pubwebapp/ and search 20250217061.