Our ability to assimilate complexity is remarkable, and this is not just limited to sight and sound. We build driving intuition, says Infineon's Chua Chee Seong.
« Previously: When will autonomous driving get in the fast lane?
Personally, I enjoy driving. Auto makers have devoted tremendous effort in making driving an enjoyable experience, for at least, when the road is clear. The frustration comes when I am in a never-ending stop-and-go driving situation in congested traffic. This is when I prefer that my car will take over the unpleasant drive. However, I have a trust issue with my car.
What has set humans apart from machines so far is our knowledge-based experience. Thanks to the experiences gained from countless hours of driving, we do not just evaluate situations on a rational level. Due to our highly developed and receptive senses, we can process a variety of situations. Our ability to assimilate complexity is remarkable, and this is not just limited to sight and sound. We build driving intuition. For example, in motor racing, the buttocks of the race driver are often referred to as the “butt meter” – providing the driver with information about the vehicle behaviour or road condition. In other words, as humans we can link our sensory perceptions with earlier experiences for a rapid and realistic assessment of the situation.
To create a viable automated car, it is important to recreate our power of reliable judgment. Here, sensors play a key role – they need to replace all of the drivers’ senses. The complexity and reliability of human perception can only be recreated technically if multiple, diverse sensors detect various aspects of the environment simultaneously.
*__Figure 3:__ High levels of automation in driving demand a significant increase in the number of sensing modules.*
Based on sensed information from camera, radar, laser and ultrasonic systems, a car can make a so-called “two out of three” decision. It means that if two out of three values coincide, this value is interpreted as being correct and is then further processed. The different characteristics of the systems increase the reliability of the machine decisions. Unlike purely optical systems, for example, radar systems function reliably in conditions of poor visibility, including snow, fog, heavy rain or glaring backlight.
Next: Can we quantify intuition? »