How do robots reproduce? Why does Volvo’s AI monitor a driver? What can Superjob network do? Our weekly digest of AI developments addresses these and other questions.
Volvo tomonitor a driver
Volvo has announced its intention to integrate an AI-fueled system into its cars to monitor intoxicated or distracted drivers.
The system will consist of several cameras capable of detecting a driver’s state judging from his\her appearance. In case of any danger signs, the auto will warn a driver by signals of the board computer. If the alarm is ignored, the neural network will alert customer support Volvo on Call as well as reduce the car speed and make it pull over.
The company plans to launch cars with embedded tracking sensors in 2020. In 2021, Volvo vehicles won’t be able to surge over 180 km/h due to the automated speed limit.
Robots reproduction
An international research group representing various universities was wondering why AI does not reproduce and decided to change the situation. The scientists are now working on the natural procreation of robots.
The research lies in an attempt to create a baby combining the parental codes. The offspring should come by two codes but develop its own one.
a bid to conduct an experiment, scientists from the Free University Amsterdam have created two robots with an algorithm tasked with reproduction. One of them was green, the other – blue. As a result, the AI-enabled robots developed their child with a combination of blue, green, and white colors.
As commented by the scientists, the algorithm uses in the field of robotics would boost creativity.
AI identifies a human profession
A service for job-seekers Superjob has presented a neural network capable of producing a human resumé by a mere photo with the accuracy of 88%.
In fact, the neural network does not identify all the parameters. It can currently recognize sex, an age, a higher education diploma, working experience, and one of 500 basic professions. What is more, AI will provide a desired salary.
In order to train the neural network, the scientists fed 10 million clothes images and 25 million resumé photos placed on Superjob to it. AI learned to correlate appearance, a salary, a career sphere, and experience.
AI to predict the useful life of batteries
The scientists from the Massachusetts Institute of Technology in co-operation with the Toyota specialists have trained AI to predict the useful life of automobile batteries.
This task is usually done by automatic control: it powers the battery and consumes the charge several times. However, its capacity is continuously growing, which requires more time for testing.
The neural network has solved the problem. Five charging-discharging cycles are enough for it to define the useful timespan. According to the authors, AI tracks data from hundreds of millions of data points and simultaneously monitors voltage.
The prediction accuracy has been estimated at 95%. In the future, this system would accelerate the development of new batteries and make them cost-effective.
L'Oréal’s AI to make cosmetics suggestions
Cosmetics company L’Oréal has developed artificial intelligence that analyzes skin and recognizes aging signs.
The neural network has been trained on the selfies of women representing Asia, Europe, and Africa. The images had different lighting.
L'Oréal eventually found the application sphere for its product. By the end of 2019, app Vichy SkinConsultAI is planned to launch. A woman will need to upload her facial image. The algorithm will sift through skin aging signs: eye wrinkles, skin elasticity, dark spots, and pore size. Based on the obtained data, the neural network will point out the zones that need care. As a result, women will get recommendations on what cosmetics to use.
Subscribe to the AI newsletter.
No spam.
