Artificial intelligence in cars has become part of a national discussion recently, with the US Department of Transportation issuing guidelines for autonomous research, Tesla’s Autopilot enabling drivers with semi-autonomous highway driving, and even Ford announcing it will mass produce fully-driverless vehicles by 2021.
But artificial intelligence in the auto world has a long history–and some of its deepest roots come from the racecar industry.
Recently, Honda R&D has been using a system, harnessing the IBM IoT for Automotive solution, to analyze data that comes from its racecars’ hybrid engines. The Watson platform, now used in F1 racing, has become key in the area of engine analytics.
“We’re using our Watson IOT platform to gather well over a couple hundred data points off of the engine in real time,” said Karen Newman, vice president of services at IBM. “Whether they’re in the trials and doing their practice rounds and different things, we get real-time data off the engine. We have an extensive dashboarding team they’re using to better understand everything from engine pressure to temperature to velocity–all those kinds of things.”
So how does it work? After the drivers perform the trial runs, the data is analyzed by a team of engineers who may, for instance, see that they need to go back and tune the engine.
Even during the races, Newman said, the data streams in through the cloud. In particular, FI racers “want high speeds and fuel efficiency,” she said. “It’s a big tradeoff game for how you tune that engine.”
“When you’re going at high speeds with all of these data points, you’re looking at all of these variables–like wind efficiency, or a wet track. I don’t think human beings can assess that amount of data as quickly,” she said, “so technologies like cognitive and artificial intelligence and analytics have always been used.”
SEE: Tech giants vs. automotive titans: The battle for your car’s data (TechRepublic)
The analytics help improve the cars after each race. A couple of weeks ago, in Singapore, the team had one of its best races, Newman said.
One reason the performance is improving, Newman said, is because the tools themselves are smarter. These cars are equipped with all kinds of sensors, picking up data on temperature, wind velocity, piston speed, pressure–and more than 270 other data points, from the engine alone.
The information is sent, through the cloud, for real-time analysis. For instance, if the temperature of the engine is too high, there may be an alert to the driver to slow down. If it’s a trial run, you could tune the engine. Temperature is associated with a lot of factors relating to the car, even down to tire wear, Newman said, which makes it an important data point to monitor.
A big reason for analyzing the data, Newman said, is to keep the cars well-maintained. “F1 has really stringent goals around fuel efficiency, different from the performance guidelines for regular consumer cars,” she said.
Newman sees the technology being extended to consumer cars. A chief engineer for Honda recently said that “a lot of things learned from racing will find their way into their regular fleet,” Newman said.
SEE: Video: IBM Watson … in less than two minutes (TechRepublic)
How do the racecar drivers, themselves, use the technology? “The drivers have headsets on, and engineers talk to the driver,” Newman said. “There’s not a lot of analytics on the racecar dashboard, because the driver needs to focus on other things–like the turn, the people next to him,” she said. “So the data analysts will say, ‘Hey, I need to lighten up on the gas in turn three,’ or whatever,” Newman said.
“When you can challenge the technology at these kinds of speeds and feeds,” said Newman, “I think it really helps us to know how well it works, so we can apply it to other less intense–under 200mph–kinds of environments, and have confidence that it will be very successful.”
Safety, Newman said, is the biggest benefit of using artificial intelligence in racecar driving. Another motivation? The bottom line.
“When you’re using these parts and moving vehicles under stressful conditions, there’s a lot of wear and tear that’s expensive,” said Newman. “These are very, very expensive pieces of equipment. If you can use the technology so that you’re not burning things out as fast, that’s value that a company can have in the long run.”
- Survey: Does your company have a use for AI or machine learning? (Tech Pro Research)
- IndyCar IT: Data protection tips from race engineers (TechRepublic)
- Tesla’s Musk says Autopilot update would have prevented fatal crash (ZDNet)
- Why the US government should take Tesla up on its offer to share Autopilot data (TechRepublic)
- Tesla’s New Autopilot Update (CBS News)