From the Mars vehicles to the autonomous vehicle and back

Nissan presented at CES the collaboration with NASA. An interview with the director of the laboratory that develops an autonomous vehicle based on NASA technology

In recent times, the people of the NASA research center have been working together with the car manufacturer Nissan on the development of autonomous driving technologies that could in the future be used by passenger cars as well as robots on Mars.

Prior to the CES show in Las Vegas last week, a group of executives from the Renault-Nissan Alliance, including Nissan Chairman and CEO Carlos Ghosn and Jose Munoz, Chairman of Nissan North America, visited the Ames Center for meetings about the technological partnership between NASA and Nissan North America. The partnership allows researchers to develop and test autonomous algorithms, chips and prototypes for a variety of transportation applications - from robots traveling on Mars to cars, and the guests also reviewed the LEAF electric car that drove autonomously through the streets of the Ames Center. The car is equipped with cameras, sensors and cellular networks and uses robotics software originally developed at Ames for robots that have explored Mars, the latest of which so far is Curiosity.

NASA says that the results of the research will be used back by NASA for the development of autonomous all-terrain vehicles for Mars.

We spoke at the Nissan booth at CES with Liam Pedersen, director of Nissan's Silicon Valley lab located near NASA's Ames Research Center: "We started working with NASA about two years ago. Several of us, myself included, are former NASA employees. For example, I managed a robotics laboratory. Nissan opened the lab in Silicon Valley because it was looking for software engineers and robotics experts. It started from personal connections.

"NASA and Nissan face the same problem. They are planning autonomous cars but in the foreseeable future, these cars will not be able to operate completely on their own and will need operators like in NASA. They have a vehicle on Mars, if it fails, you can't get there and fix it. The robot itself knows how to take care of itself. He sees obstacles and bypasses them but he cannot tell what is important and what is not. He cannot do the science by himself and does not know how to deal with unusual situations."

"NASA solves the problem thanks to a team of operators who work in front of the robot with a huge delay - it takes the signals to reach Mars and back about 40 minutes. Therefore they give him instructions to get from one point to another and to carry out the tasks planned for him along the way. The robot has to handle problems by itself and react to sudden events."

"The same problem also exists with autonomous cars. We can already today provide such a vehicle with basic safety, avoid collisions and bypass obstacles, but what is difficult for a machine is to interact with humans. If someone makes a sign with their hand (for example asking to open the window) it is not something that a machine can interpret. Especially if the movements weren't really clear. Work zones also pile up difficulties for autonomous cars."

"The idea is to manage fleets of robotic cars, each car maintains its safety, and drives itself 98-99% of the time, maybe even 99.9999% of the time without involving a human, but there will always be cases where the ability of human intelligence will be needed. Example - a person directs the traffic at an intersection, the traffic light says to stop, and the traffic policeman says to continue. And now imagine a fleet of a million cars. Even if they are fully autonomous, sooner or later they may get stuck. That is why we, together with NASA, are developing a system of remote operation for managing fleets." Petersen says

"It is clear that artificial intelligence will be used so that the car can manage on its own as much as possible, it is also economically viable because it is desirable that one person be able to monitor as many cars as possible and if the cars are not completely autonomous and need help frequently it is not worthwhile.

We want to take care of the extreme cases. Anyone who understands machine learning will tell you that it is very easy to develop a machine that can manage on its own 99 percent of the time, but when extreme cases arise it is very difficult because there are thousands of problems and each one requires its own unique expertise.

How soon will it be possible to see smart fleets and what will be the impact of this transition on society, in particular in the field of employment?

"This is a business question. From a technological point of view, we will reach this in three years. But when exactly they will be on the road - Nissan will have to decide on that with its customers." Pedersen says.

"I live in San Francisco, my office is in Sunnyville. I spend an hour every day in each direction in the car. I would be happy if I could take advantage of this time, for example to sleep, go through emails, and do something.' Today is a time that is being lost. An autonomous car will give me two extra hours every day. Besides the issue of time, safety will be an important issue, especially in private vehicles."

"Many companies are looking into the field of autonomous driving, because there is a lack of drivers, especially in the field of logistics. I assume that in the first phase the autonomous capacity will help to fulfill the demand, but when they become more common I predict a significant change in our society."