Phoenix Self-Driving Tests Show Need for Predictive Maintenance in Extreme Conditions

Movimento Press Blog

predictive maintenance

(Image Credit: Wikimedia Commons User Simeon87 )

Chances are, you have never driven in the Dakar Rally, a punishing car race that used to take place in the African desert and is now held in equally difficult South American climates. The race is grueling not just due to its length, or the rough and cruel terrain, but because the heat and sand are terrible for engines. The Dakar Rally has long been a mixture of driver brilliance and mechanical ingenuity. Drivers must anticipate everything that can go wrong and react perfectly when something does. If you were to take away the actual driver, you can see how much more difficult the race would be.

That is the theory behind the testing of self-driving cars in extreme conditions. Autonomous cars will not only have to adapt to these inhospitable climates, they will have to do so without the benefit of drivers who can notice when something feels off. For the cars, this does not just mean reacting: it means predicting what can go wrong in any given condition. Meeting this challenge will require secure and rapid OTA communication so that OEMs can not only predict problems but preemptively avoid them, keeping cars on the road and passengers safe, whether in the deserts of the American Southwest or the blizzard-filled cold of a North Country winter.

Google Tests Self-Driving Cars in Phoenix

It is hot in Phoenix. If you have never been there, it is America’s hottest major city, with temperatures frequently reaching into the triple digits. This heat is magnified by older infrastructure, including vast asphalt parking lots that trap the sun’s relentless power, making the city even warmer. So it stands to reason that when exploring extreme conditions, Google chose Phoenix as a test site for their self-driving cars. As Jennifer Haroon, the head of Business Operations on the Self-Driving Car Project told C|Net, “The Phoenix area has distinct desert conditions, which will help us better understand how our sensors and cars handle extreme temperatures and dust in the air…. Driving in new cities enables our engineers to further refine our software and adapt to these different environments.”

The other cities are Austin, and Kirkland, Washington, where it is considerably rainier than Mountain View (to say nothing of Phoenix). Google is adamant about testing under these conditions because cars do not do very well with heat and dust, or with excessive rain.

There are two reasons for Google’s tests, encompassing different but overlapping features:

  • To ensure the self-driving car’s ability to communicate difficulties through its algorithm and to receive transmissions while operating under extreme conditions. This functionality is enormously important for OEMs, because self-driving cars will not work without a steady stream of OTA updates.
  • To compile data about what can go wrong in every condition. Without a driver, cars will have to be able to anticipate what is going wrong, or what could go wrong. Testing in these conditions allows OEMs to establish a crucial predictive maintenance regime.
The Importance of Predictive Maintenance and OTA in Harsh Climates

Predictive maintenance is knowing when something can go wrong and fixing it before it does. In most modern cars, simple cases are already done today. Perhaps a car senses it is running out of oil and a light goes off, allowing its owner to take action before the engine dies. It is a routine everyone is familiar with, but easy as it is, it still requires interaction between the car and a human driver.

With autonomous, software-driven cars, this process is trickier. Predictive maintenance requires a connected device to monitor and judge when it needs fixing and let the human owner or OEM know. This encompasses more than just an oil change. This requires sifting through terabytes of Big Data and organizing that data in a way that allows patterns to form algorithmically. This allows the car to “know” when it is facing conditions where something can go wrong or lead to a breakdown.

Predictive maintenance is essentially creating a set of rules that a car’s algorithm can follow and interpret. Here is an example. If there is a dust storm, a haboob, swirling up from the desert, then other cars slow down, visibility is lessened, and software can malfunction. Therefore, when there is a dust storm, the car will “sense” that it needs to stop driving so it does not get damaged. It knows that it has to take certain precautions.

Predictive maintenance is also important in non-extreme conditions and extends beyond automotive into the Internet of Things (IoT). The heart of the IoT is this sort of predictive and adaptive repair work; the IoT cannot function without it. Devices, from huge turbines to car airbags to homes toasters, have to be remotely managed, fixable without sending someone all over the country to do so, and able to “think” for themselves and request and receive upgrades. In any industry – from automotive to extraction – having devices that can essentially fend for themselves is an economic and pragmatic requirement.

This means that over-the-air communication and Big Data collection/analysis are imperative. Intelligent communication will be what makes self-driving cars work, with the ability to collect data and implement it, to share that data with the entire network, and to assimilate new information and SOTA/FOTA updates securely. All this will keep passengers safe and driving efficient. Very few drives are as rough as the Dakar Rally, but every drive collects and uses millions of data points, and without OTA communication, successful self-driving cars will be lost in the desert.

Subscribe to our newsletter