News of the week selected by Impactscool – June 22nd

To save energy, drones could take the bus

All over the world, experimentation projects are underway for the delivery of goods to home via drones. One of the problems that most grips these systems is the limited range of action. A good drone has a range of about 20 km before its battery runs out. Research from Stanford University has speculated that a possible solution would be to get drones to take public transportation. The idea, at first sight bizarre, is actually very sensible, developing a system that allows drones to land on the roofs of buses and get a ride would allow them to significantly expand their range of action without substantial changes to the system power supply.

Now you can buy a Boston Dynamics robot for your company

The robotics company that in recent years has amazed us with demonstrative videos of extremely agile robots has officially opened the sales of its SpotMini robots. Small quadrupeds have many applications, as shown in the accompanying video, and for now they are only available to American companies, from next year sales will open to the international market. The cost, definitely not accessible to all: 74 thousand dollars for the basic model, to which a series of options can be added such as remote sensing sensor systems, extra computational units, omnidirectional cameras and more.

This AI generates faces from simple sketches

Researchers from the Chinese Academy of Sciences and the City University of Hong Kong have developed artificial intelligence capable of generating realistic faces from simple sketches. DeepFaceDrawing, this is the name of the artificial intelligence algorithm, allows users to create incredibly realistic portraits by inserting simple non-professional sketches, roughly drawn. The program is based on machine learning to fill in the spaces between the various basic parts such as eyes, nose and mouth. A huge database of facial features provides detailed information about the filling which therefore generates a realistic face. DeepFaceDrawing has great potential for creating portraits of suspects but also in applications of facial recognition, medical procedures and educational training.

The first artificial synapse capable of “learning”

A research published in Nature describes the first artificial biohybrid synapse capable of interfacing with neurons in an active way. The result, the result of the efforts of researchers from the Italian Institute of Technology (IIT), Stanford University and the Technical University of Eindhoven, is an exceptional record that opens the way towards a better connection between man and machines , new tools to treat degenerative diseases and new developments for neural networks. The synapses take care of transporting the electrochemical signals between two neurons, in the research the signal starts from a group of mouse neurons capable of transmitting the dopamine neurotransmitter and arrives at a microchip capable of activating – like a real neuron – after reception of a certain amount of dopamine, and, above all, to take into account the received dopamine.

There is still a long way to go, the artificial synapses are still too large to find real use, but research is proceeding towards miniaturization and the prospects are many.