Self-driving cars may be the future, but having a driver in the loop will likely be around for a long time. Driver monitoring systems (DMS) can help improve safety by identifying states such as a sleepy driver so that the system may provide stimulus to prevent an accident. This type of system can also be use to monitor passengers in a robo-taxis scenario.
Original equipment manufacturers (OEMs) and Tier 1 suppliers are looking to build advanced DMS support as part of their advanced driver assistance systems (ADAS). Affectiva is delivering a system that utilizes cameras and microphones to track these status of the driver and passengers (see figure).
Affectiva uses the term artificial emotional intelligence, or Emotion AI, to describe its emotion recognition and emotion detection technology. The company uses speech and facial encoding to provide a richer insight into the human expression observed using cameras and microphones. Deep learning algorithms are used to analyze images to classify facial expressions. The system measures seven emotion metrics, including anger, contempt, disgust, fear, joy, sadness and surprise. It also provides 20 facial expression metrics.
Affectiva’s system can track the status of all passengers in a vehicle.
“Affectiva is the only AI company that can deliver people analytics using algorithms that are built with custom-developed deep learning architectures,” said Dr. Rana el Kaliouby, the company’s CEO and co-founder. “We have built industry-leading Emotion AI, using our database of more than 6 million faces analyzed in 87 countries. We are now training on large amounts of naturalistic driver and passenger data that Affectiva has collected to ensure our models perform accurately in real-world automotive environments.
“With a deep understanding of the emotional and cognitive states of people in a vehicle,” el Kaliouby continued, “our technology will not only help save lives and improve the overall transportation experience, but accelerate the commercial use of semi-autonomous and autonomous vehicles.”
The first release of Affectiva Automotive AI supports in-cabin face and head tracking of all occupants simultaneously. It supports multiple camera positions and can take advantage of near-infrared and RGB cameras. The software runs on NVIDIA hardware and will be demonstrated at this year’s NVIDIA GPU Technology Conference.
Affectiva’s technology is available for use in other applications. Its Affdex SDK can be used to analyze video input and generate facial expression output in the form of Affdex metrics. These include the seven emotion metrics and 20 facial expression metrics already mentioned, as well as 13 emojis and four appearance metrics. The SDK can also be used to measure valence and engagement.
