Your Drive of the Future

As you drive down the road, the person in your passenger seat screams, “Watch out!” Will the next moment be a life-altering event? Or will you drive on safely because you, as the driver, are prepared for the pedestrian that suddenly stepped into your lane from behind a stopped bus? We are working to prepare you for and eliminate the “Watch out!” events. How? By developing applications for smart computers and cameras that “look ahead” at the road. These applications will let you, as the oncoming driver, know of nearby pedestrians, cyclists, and vehicles hidden from your view.

Pedestrian traffic fatalities in the United States continue to rise. The Governor’s Highway Safety Association reported that 2018 and 2019 were the deadliest years for pedestrians since 1990. This rise in pedestrian traffic fatalities is a problem that can be solved with the help of applications on roadside computers that communicate with vehicles (known as V2X or C-V2X). sibrtech inc. is developing “look-ahead” technology for V2X / C-V2X applications to give vehicle drivers advanced “sight” of hidden or unseen hazards and vulnerable road users. “Look ahead” is a key component of advancing road safety and it works for both human and automated drivers.

The best way to explain how it works is to put you into an example from our testing.

On December 9th, 2020, sibrtech completed a third day of testing at the American Center for Mobility of this very technology. The American Center for Mobility is a closed course testing facility in Ypsilanti, Michigan for autonomous vehicles and other advanced transportation technology. We used an inexpensive artificial intelligence (AI) computer and a camera in a box eight inches long, five inches wide, and three inches tall mounted almost 30 feet high on a light pole next to a six-lane intersection. The components of this box cost less than three hundred dollars. The AI computer ran our software that identified road users–either one or more vehicles or persons, determined their location and approximate size, and transmitted this data to an oncoming vehicle. Our software identified road users 345,000 times in a little over 3 hours running at about 20 frames per second.

Here is a picture of the camera and AI computer mounted on the light pole on December 9th, 2020.

Figure 1. sibrtech camera and AI computer mounted on a traffic light pole at the American Center for Mobility.

Your drive of the future

You are once again the driver. The photo below — a video frame from our testing — is your view approaching an apparently empty intersection. The shipping container in the right turn lane simulates a double-parked truck. You are planning to turn right at the intersection. Your view of the upcoming right turn path is blocked by the simulated parked truck. This blocked view may cause slight anxiety because you cannot see your turn path. Or the view may lull you into a sense of comfort because, after all, the intersection looks empty.

Figure 2

Briefly let us take you to our AI computer and camera. The next figure below is a view of the same intersection from the sibrtech camera and AI computer mounted on a light pole on the far side of the intersection. This is a big intersection. The camera has a view of approximately half of the intersection and of the north/south road approaching the intersection. This is our computer’s view while you, as the driver in Figure 2, approach the intersection (and by now you are getting a better understanding of what we are talking about). If you are looking for your car, it is largely hidden by the container acting as the simulated truck in the right turn lane. Our software running on the AI computer mounted with the camera on the light pole identifies six persons in the intersection hidden from your in-car view in Figure 2. It also identifies another vehicle approaching the intersection. The other vehicle is the same one that you see on the left in Figure 2. If you were the anxious one, you win the day. What appeared as an empty intersection is in fact a populated intersection.

Figure 3

Now you are back in your vehicle approaching the intersection. Only this vehicle has something your everyday car does not — it has a computer that receives wireless data from our AI computer on the light pole. The AI computer on the light pole identified the locations of the hidden pedestrians and sent this information (along with other data) to your in-vehicle computer.

For purposes of this test, we created a simple app to run on a computer in your vehicle. This app receives the data transmitted from the AI computer and shows the location of the hidden pedestrians. This simple app is not a final design for an in-vehicle system, we will leave that to the car companies. A screenshot from the app taken during the test is shown in Figure 4, on the right. The app draws a red box around pedestrians closest to your path and a yellow box around pedestrians near your path (the green box is the other car).

Figure 4. On the left the in-vehicle view; on the right, a simple in-vehicle visual alert of two groups of hidden pedestrians.

The two groups of pedestrians are clearly visible through the in-vehicle app, illustrating the ease of alerting you to hidden pedestrians. You have no reason to be anxious after all. You can plan the right turn maneuver knowing that multiple pedestrians are in or near the right turn path. This is true even though you would not actually see those pedestrians with your own eyes until passing the obstructing truck.

To further that point, here is what you see having just passed the edge of the truck. You are not surprised to see one group of pedestrians about to cross right in front of you and another group in the crosswalk approaching your right turn path. “Look ahead” already told you that they were there.

Figure 5. Your first actual view of the pedestrians.

The image below helps illustrate that the AI camera determines positions of the road users in three-dimensional space. This position information is useful for any vehicle approaching the intersection from any direction.

Figure 6. Detected road users in three-dimensional space.

Our technology approach is easy to get up and running on roads. For example, we processed the data on a computer mounted with the camera. This eliminates the need to transmit video elsewhere for processing on a “bigger” computer. With power, a low bandwidth connection for remote configuring, and a communication link to the vehicle, a system can be operational. A driver in any vehicle equipped to receive the communications can take advantage of the data and be part of the next step in transportation safety. It works for vehicles with human drivers and for autonomous vehicles. Our unit was inexpensive. That means infrastructure planners need not make a choice between fixing bridges and roads or adding safety technology. They can choose both. The technology to make “Watch out!” events safer can be done.

We want to thank and are extremely grateful to the PlanetM initiative of the Michigan Economic Development Corporation for a grant making our testing possible.

Finally, thank you to the team at the American Center for Mobility for the great support during our testing.

If you would like to learn more, please contact sibrtech at info@sibrtech.com. sibrtech inc. is a Michigan-based tech startup focusing on edge computing and transportation safety.