Driverless Yoyo. An amazing story
That is, I’ll tell you about the time we equipped a mini-car yoyo with artificial intelligence.
It – could – work!
After years of research and design, for clients or investment proof of concepts, we decided to create a prototype to answer this question: can we turn a vehicle into a smart car?
Our partnership with XEV allowed us to create this prototype based on the Yoyo electric city car. But what did we have to do? What do a vehicle neeed to see, perceive, decide while it is on the road amidst a thousand variables? How does technology allow us to replace human drivers by reducing errors and dangers?
In this journal we tell our story. The story of a car that one day, in Teoresi, we equipped with artificial intelligence.
The state of the art of autonomous driving technology on wheels takes us into a fascinating world: information gathering is done through HD cameras, lidar, ultrasound, radar, GPS integrated in real time through sensor fusion technique giving the in-vehicle computer brain a complete picture that allows it to digitally reconstruct the surrounding environment.
We remember the day we mounted and tested lidar and cameras on the XEV Yoyo very well. The 3D printed bar did not fit properly, we had to make adjustments, then we turned it on and on our computer screen we saw in real time what its digital eyes were seeing.
How could we test the functionalities of our project before moving on to physical modifications and road tests? By training our artificial intelligence software in a virtual environment, as in a video game.
We first performed a 3D scan of the vehicle and then created a meta-environment where artificial intelligence makes virtual object, pedestrians and cars move, reconstructing the randomness of the real world. The system is trained to avoid obstacles, recognize and avoid pedestrians, start the sensor fusion that the physical car will then have, and execute a path while staying on the track.
Drive-by-wire is how transportation engineering defines all automated control systems in a vehicle. Typically brake, accelerator, steering, and other controls are controlled by electronic control units using electromechanical actuators. In order to give our artificial intelligence total control of our Yoyo, we had to make all the controls that are usually in human hands electronically controllable. It was a work of system integration, but also often of a lot of patience with drills, files, saws, because here the mechanical, electrical, and electronic parts we added have to match perfectly with the original ones.
The time has come to put into action all that we have developed in the artificial world.
We have to test whether the car, just as it recognized obstacles and pedestrians in the 3D virtual environment, can now do it accurately in the real world as well. It’s true, we’ve already done it on other prototypes, on scale vehicles, we know our garage very well, the surrounding private path, and we’ve made our system know it very well. But will it know how to recognize a human being crossing the street? Early tests with stationary car are great! Now we have to really try, safely, limiting the speed of the car to the minimum.
It works! The algorithms are stable, even in different light conditions. Our Yoyo now brakes by itself if it finds an unexpected obstacle in front of it, such as a pedestrian.
Today is an important day. It is the day.
Our entire design and development team is assembled, colleagues have come from all locations: Naples, Turin, Detroit. About 40 colleagues have been invited to the in-house presentation. Everything has to work at its best. We are sure that it will, but the demo effect is being felt, the excitement is only felt by looking into our eyes, peeking out from the face mask.
Our XEV Yoyo sets off, with a passenger on board. There is a strange feeling in observing the empty driver’s seat. It takes a predetermined route, perfectly maintaining the roadway, stops when a pedestrian suddenly crosses the road, makes a left turn, avoids an obstacle, and parks in front of the garage entrance. Ovation. We did it.
Torino, l’auto del futuro e i test di guida autonoma alla Teoresi
Uno sguardo nel futuro dell’automobile: smart e connessa
Frena davanti ai pedoni e svolta agli incroci, Teoresi porta la sua city car a guida autonoma da Torino a Milano
Guida autonoma, Torino è capitale d’Italia: ecco i benefici di viaggiare senza conducente
Dalla sperimentazione di guida autonoma al Centro Nazionale per la Mobilità Sostenibile: Teoresi cresce nella smart mobility
Cybersecurity on the move, that is, travelling safely with the cars of the future
Cars: Teoresi and Torino City Lab for the autonomous driving