What’s the difference between a human and a machine?
A human has a brain that can process millions, if not billions, of calculations per week. In tandem with the spinal cord, every step we take is calculated in real-time, and the signals are sent to our limbs for action. Imagine the amount of passive processing our brain does for simple tasks like getting up from a chair.
Similarly, driverless cars have to address similar challenges. With its array of sensors on all sides, it’s continuously processing information to understand the situation and take actionable measures. We’ve seen numerous videos of Tesla’s AutoPilot system kicking in automatically to prevent a crash.
Though, these instantaneous reactions are based on directly available obstacles. While we are driving, humans acquire “judgment”, it generally means the amount of experience we have. The more we drive, the more our brain knows how to react in various situations. Parallelly, our brain also learns to predict paths and outcomes based on factors like direction, speed, and other external causes.
Predicting human behavior is no easy task, especially when the car is essentially a “machine” that depends on human-derived data sets. And, we need this property because computers cannot be expected to perform only on real-time data.
This ability to predict and respond is crucial for a driverless car. And, a Berlin based German start-up called Phantasma Lab has received US$1.2 million in 2018 as funding. The company is trying to solve the problem by building human behavior data sets that computers understand and process.
Using virtual simulations based on mathematical rules, the company intends to further research and improve the accuracy of these data sets. It is already selling them to driverless car companies and generating revenues.
Maria, CTO of the company, told SCMP, “Sensors will help the car recognise what was there, but not what to do after that.”
Phantasma is targeting China as its prime market because the country already has a number of companies that are working on driverless technology. This includes Baidu, Pony.ai, and Auto X.
Perceptive Automata is an American company that’s also working on addressing similar issues. They are researching behavioral science techniques to characterize the way human drivers understand the states of the minds of other humans.
Undertstanding human behavior will be crucial for driverless cars because south-east Asian markets are filled to the brim with challenegs. In my previous article I’ve detailed how India is not a good match for driverless cars right now because traffic laws are barely followed. In cases like these, the system needs to be capable enough to predict these in advance and act accordingly when required.