Within the enormous casuistry to be tackled by artificial intelligence in driving is dealing with extreme situations where the important thing is to preserve the integrity of the vehicle and the passengers. The discipline of drifting can be very useful to maintain control in certain unexpected situations.
A year ago the Toyota Research Institute presented us with a prototype of the GR Supra with an obvious widening of tracks and various modifications to improve the practice of drifting. But the underlying idea was to make a car that could do drifting without any driver at the wheel, a human driver.
As a reminder, the drifting is a motorsport discipline in which it is about linking curves at the maximum possible speed while keeping a delicate balance between tire grip, speed, rotation angle, etc. It is not usually within the reach of any common driver.
And it is that the techniques used in the world of competition have their practical utility for everyday driving, especially when it comes to dealing with very complicated situations that require skill and vehicle control. These techniques can prevent road accidents when used properly and at the right time.
Well, we have news about that Toyota prototype, which is the world’s first autonomous car in the world of drifting. To begin with, it has feline reflexes, trajectories are calculated 20 times per second, faster than an expert driver in the discipline. Professional driver Ken Gushi collaborated with Toyota engineers.
At the moment the prototype is not completely autonomous because you need to know in advance where the obstacles are and is not able to recognize them in any other way. In fact, the car is only “known” for a 3.2 kilometer track at Thunderhill Raceway Park (California, USA).
When it comes to avoiding a sudden situation of lack of grip, such as stepping on an oil stain, gravel or ice, we have two options, avoid them by going around them, or go through them trying to maintain control. That is what it is about teaching artificial intelligence, regain control of the situation instead of just trying to stop it with as little damage as possible.
Avinash Balachandran, Senior Manager of TR’s Human Centered Driving Research Team, said: “We are expanding the range in which a car can be controlled, with the aim of giving ordinary drivers the instinctive reflexes of a professional racing driver to be able to handle the most challenging emergencies and keep people safer on the road.”.
Also collaborating in the project are the guys from the Dynamic Design Laboratory at Stanford University, who previously demonstrated the potential of this technology with a modified DeLoream DMC-12 with electric propulsion. We were able to see the result in the famous video entitled «MARTYkhana». With Toyota they have gone further.
The practical applications of this technology include both assisted driving with people behind the wheel, and artificial intelligence driving autonomous cars. It is certainly a relief to think that autonomous cars will be able to deal with very dangerous situations such as getting out of a skid or, failing that, that the smack is much milder.
Nevertheless, this technology is not intended to replace the driver, but to improve it and “magnify” it with help that can be essential in an emergency. It doesn’t look like Toyotas will have a “mode” in the future. drift» that is used incorrectly – you all know what I mean.
At a later stage of development, the car will need a series of sensors to be able to detect an emergency situation earlier. As for the steering wheel, there is an interesting technical debate, should the steering be temporarily uncoupled so that the driver does not hinder the correction, or will he have to have the cold blood to partially release it for the artificial intelligence to act, as test drivers do? ?
At the moment car designers they always give the driver the ultimate power of decision. For example, if a lane keeping assistant corrects us not to go out, we can always ignore the correction and force the lane departure. It will be more or less stupid, but the machine does not make the final decision.