Robots see through the walls, the military use Microsoft helmets and other news from the world of AI

We strive to inform readers not only about events in the Bitcoin industry, but also to talk about what is happening in related technological spheres-cybersecurity, and now in the world of artificial intelligence (AI).

Forklog has collected the most important news from the world of AI over the past week.

• ARM introduced a new generation of processors with support for artificial intelligence – for the first time in ten years.
• Microsoft won a contract for the supply of Hololens mixed reality helmets for the US Army. The agreement was estimated at almost $ 22 billion.
• Researchers from the Massachusetts Institute of Technology have developed a robot that can “see” through the walls.
• A group of scientists from the University of Toronto and LG AI Group has developed an “explained” artificial intelligence to eliminate the problem of the “black box”.
• The number of video cameras with artificial intelligence in smart cities will reach 350 million by 2025.

ARM introduced the ninth generation of processors with AI support

ARM chips developers have announced ARMV9 – the first architecture of mobile processors over the past ten years. In the new generation, manufacturers concentrated efforts on three main improvements: safety, artificial intelligence and higher performance in general.

ARMV9 will support Scalable Vector Extension 2 (SVE2), which is designed for machine learning and digital signal processing tasks. SVE2 should ensure a significant increase in performance for virtual and augmented reality, as well as image processing and voice recognition.

The new generation of processors will support the Confidential Compute Architecture technology – a hardware environment that is created dynamically and protects the important data of users from the rest of the system.

ARM also promises a general increase in the performance of new architecture by more than 30% due to optimization of software and hardware. The company assured that all programs created for previous architectures will easily start on ARMV9.

Partners of the company can expect shipments of the first parties until the end of the year.

Boston Dynamics introduced a robot to automate warehouses

The manufacturer of Boston Dynamics robots showed the new Stretch robot to move the boxes to automate warehouses and distribution centers.

Boston Dynamics Is Bringing Mobility to Warehouse Automation. Watch Stretch – OUR New Case Handling Robot – Move, Groove and Unload Trucks.

The robot has a square mobile base on wheels, which allows it to move through loading docks, maneuver in a limited space and adapt to the changing layout of the object.

To move the boxes, Stretch uses a large manipulator with suckers with which it fixes the load.

The company also stated that the robot is equipped with branded computer vision technology Boston Dynamics. With its help, Stretch can recognize different boxes without the need to retrain the model for each specific user.

In one hour, the robot can move 800 boxes weighing up to 23 kg each. Large containers allow him to work eight hours without recharging. To manage the robot, the operator does not need special training.

At the moment, the company is looking for partners who will take part in pilot testing of robots. The developers did not report how much Stretch will cost, but they plan to launch sales of the device in 2022.

Microsoft will supply US Army 120,000 supplemented reality helmets for $ 22 billion

Microsoft won the tender of the US Department of Defense for the supply of augmented reality helmet. A contract, a potential cost of $ 21.88 billion, provides for the shipment of 120,000 devices over the next ten years developed on the basis of Hololens technology.

This is not the first contract between Microsoft and the Ministry of Defense. In 2018, the company and the department signed a contract for $ 480 million, in the framework of which the specialists of the software giant were supposed to put modified mixed reality helmets specifically for the needs of the military.

Improved sets combine various sensors, the information from which is projected into the built -in display.

Then some employees protested and signed an open letter to the company’s management not to sell technologies to the military.

General Director of Satya Valel answered critics, but this did not prevent the company and the department jointly work on the updated set.

Researchers created a robot that sees through the walls

Researchers of the Massachusetts Technological Institute created a robot that can see objects through obstacles.

The developers complemented the classical computer vision system of radio frequency identification technology (RF), the signal of which is able to penetrate even through the walls.

Scientists believe that their development will help reduce the bias of algorithms. This is especially important in sectors such as medicine, where inaccurate information can cost the life of the patient.

“In order to fully understand and interpret the prediction of the model, the doctor should know how the algorithm came to him,” added Mahesh Sudhakar, a graduate of the University of Toronto and one of the members of the research group.

Research: The number of AI cameras in smart cities will reach 350 million by 2025

In 2025, 350 million cameras with artificial intelligence will work in smart cities around the world, according to the study of the analytical company Abi Research.

Experts are sure that by this time the use of AI chipsets in the cells will become the norm, and the deep training algorithms launched on devices will help urban and law enforcement services to drive public transport, observe pedestrians, detect threats and ensure security.

They also highlighted two technologies that will become catalysts for the development of AI in cameras in the coming years.

The first is the widespread use of Edge devices that will allow you to process video data in the places of their generation. This will reduce the load on communication channels, because the devices will not need to be sent to the server gigabytes of video data for their subsequent processing.

The second technology noted by experts is 5G networks. They noted that telecom operators will not have time to put the fifth -generation network into the commercial operation of the network in the coming years. However, their point deployment will ensure a reliable and seamless connection of urban infrastructure, which is precisely necessary for machine learning.

Analysts do not exclude that integrators of smart cameras may encounter difficulties when turning total control devices in cities. First of all, this is resistance from the public and human rights activists who already express concerns about the transparency of computer vision technologies, especially persons recognition.