As the push to autonomous driving continues, so too do the debates around which kind of sensors will provide the basis for eventual success. While there are many proponents of radar, lidar and camera sensors, there is a growing consensus that a combination, or fusion, of sensor technologies will ultimately win out. In this article we look at how our VIA Mobile360 ADAS (Advanced Driver Assistance System) solution can be enhanced with the addition of radar sensors to improve the accuracy of Forward Collision Warnings – a fusion of sensor data that we refer to as VIA Mobile360 Sense.
Adding Radar Sensors to VIA Mobile360 ADAS
VIA Mobile360 ADAS solutions typically include a mix of several standard safety features including Lane Departure Warning, Front and Rear Collision Warnings and Blind Spot Detection. As well as adding other features such as Dynamic Moving Object Detection and Driver Monitoring for example, we’re also constantly updating and improving the AI algorithms that make it all work at the software level. One other way in which we can improve the accuracy of our system solutions is by using VIA Mobile360 Sense, a technology which fuses data from different types of sensors.
Our standard VIA Mobile360 ADAS system uses FOV-40° automotive grade cameras connected to a ruggedized computer vision system. However we have recently started to augment our ADAS solution with radar, a sensor technology which should bring some very strong advantages to the table.
Radar is obviously a mature technology that has been around since before the Second World War. It works by firing radio waves at a target and measuring the frequency of the wave’s reflections to calculate the distance and velocity of the target object. It’s an accurate way for a vehicle to detect other vehicles, pedestrians or objects on the road, assessing how far away they are, and how fast they are moving – the perfect data required for accurate Forward Collision Warnings (FCW).
The problem with radar is that it tends to detect everything within its specific field-of-view or range, including trees, buildings and even blades of grass adjacent to the road. This is where VIA Mobile360 Sense really comes to the fore. The FOV-40° camera mounted on the front of the vehicle uses object detection algorithms to help the radar to focus only on objects that matter (cars, trucks, pedestrians etc.). The cameras detect the target object and allow the radar to then accurately calculate distance and velocity.
In the video below we have a demo of VIA Mobile360 Sense in action. The video screen on the upper right shows the road ahead, with the camera using object detection to detect other vehicles on the freeway. Detected vehicles are identified with a red box. Radar targeting is represented by a small blue circle, while distance and velocity are displayed in green in real-time.
Note: The left side of the screen has been set up to show the raw data being fed into the computer. Our car is shown as a small red square at the bottom of the display.
It’s also worth noting that the cameras are using AI algorithms to detect the lane that the vehicle is driving in (i.e. the two blue lines in the data feed). This allows the system to employ Lane Departure Warnings (LDW) that alert the driver if he or she strays out of the lane, due to tiredness or distraction for example.
The demo in the video uses a single radar sensor mounted above the bumper on the front of the car. The sensor can be configured in short range mode which offers effective detection at distances of up to 60 meters using a field of view of 90°. Alternatively, the radar can be configured in long range mode with detection possible up to a distance of 127 meters using a narrower 20° FOV.
This demo is an example of how VIA Mobile360 Sense fuses data from two different types of sensor technology in real-time to give the system a clearer picture of the road ahead. It’s the kind of technology that autonomous vehicles will ultimately require, and can be seen as a crucial step in that direction. For now however, it’s an important step towards eliminating road accidents and saving lives.