Artificial intelligence (AI) is being used in the automotive space to anticipate and understand unpredictable human behaviour as well as to create intelligent driving functions.
To celebrate arguably the most important shift the automotive industry has seen since its creation, Auto Futures is launching “AI and You,” a special report that will offer in-depth coverage of the emerging sector over the coming weeks from leading AI companies and analysts.
In a preview, Auto Futures speaks to three AI experts championing the technology.
“Automotive AI uses a combination of techniques, including machine learning (ML), deep learning (DL), natural language processing (NLP) and computer vision (CV) to identify patterns in sensors or operational data, and then infers actions based on a set of predetermined rules for operation,” says Clint Wheelock, Managing Director of AI research firm Tractica.
Affectiva’s Emotion AI technology uses AI to figure out what people are thinking and feeling. Affectiva started out of MIT in 2009 and analyses seven million factors and a wide breadth of data, from all over the world.
“Affectiva is very good for vehicle safety. Systems will be able to respond in real-time to help prevent accidents,’’ says Ashley McManus, Affectiva’s Director of Marketing. “Affectiva can be used for driver state monitoring. In the future, it can be used to monitor stress.”
McManus says that Affectiva is currently working with seven major automakers who will decide how to warn passengers when a safety issue is detected. It’s also working with Nuance’s Dragon Drive, deploying its multi-modal in-cabin AI sensing solution for an interactive automotive assistant that understands drivers’ and passengers’ cognitive and emotional states from face and voice data.
Sentiance, a Belgian sensor profiling startup, provides an AI platform that analyses data to create a deep understanding of human behaviour and context to build new and personalised products and services in domains such as health, mobility, smart living and commerce. For mobility, the company uses smartphone data that includes GPS, accelerometers and gyroscopes which follow the user around and extrapolate what the user or driver is going to do.
“It’s all about personalisation of the car experience to make peoples’ lives easier,” says the President of Sentiance, Dimitri Maex.
Sentiance technology is used within Peugeot’s Instinct Concept vehicle. Depending on the purpose of the drive, such as shopping, commuting or dropping off the kids, the car is able to automatically change its characteristics, such as lights, position of the steering wheel and seating position.
Maex says the company’s algorithm is so advanced that it was even able to figure out that an employee was going to the gym on a day that he usually didn’t, due to coming to work early and leaving early.
Sentiance is currently working with five automakers and Maex sees the possibility that this kind of data can be used for things such as pre-conditioning the car in inclement weather, automatically setting the destination in the mapping system and for mobility predictions such as knowing when someone will need a specific vehicle.
AI Maintains, Connects and Drives Revenue
Other use cases for automotive AI include predictive maintenance, connected vehicles for intelligent infrastructure and autonomous vehicles, says Tractica’s Wheelock.
An example of predictive maintenance is CarFit, which uses machine learning by scrubbing data of noise, vibration and harshness to predict mechanical issues before they become a problem.
Connected vehicle AI deploys hardware, software, and services designed to support vehicle-to-vehicle and vehicle-to-infrastructure communication. The goal is to allow a safer, more responsive and more intelligent transportation system, says Wheelock.
For autonomous driving, AI uses computer vision (sensors, radar, mapping and LiDAR) with deep learning for navigation and object detection.
“Because these systems do not suffer from human emotions, they are not subject to making rash decisions based on anger, fear, or other conditions that can impair decision making,” says Wheelock.
“Another very important key to autonomous driving is building generative models of the real world,” said Wheelock, “This approach tests autonomous vehicles inside virtual worlds that include accurate models of highways, roads and environments.” Uses of automotive AI will continue to grow and generate revenue. Tractica estimates that it will generate $26.5 billion in annual global revenue by 2025.