Thus the «robodog» will help to find (and rescue) the victims of an earthquake – breaking latest news

He pawed through the rubble of a building partially destroyed by an earthquake. She ducked into pipes and tunnels looking for any survivors. For the first time Spot, Rice’s bionic dog (Robotics and Intelligent systems for Citizens and Environment) Dibris (Department of Computer Science, Bioengineering, Robotics and Systems Engineering), University of Genoa, tested its capabilities in search and rescue operations.

Let’s be clear: it was a simulationcarried out today on the training ground for dogs (in the flesh ) that Anpas inaugurated in July 2022 in the Civil Protection area in Foligno (Perugia). A preliminary analysis to acquire a series of information, as explained by Antonio Sgorbissa, associate professor of Robotics in charge of Rice. The USARBOT project – Robotic technologies for search and rescue in serious environmental disasters involves, in addition to Rice Dibris and Anpas, also the Degree Course in Nursing – Terni Campus – University of Perugia and the Politecnico di Milano.

The Dionysus project: development of innovative systems for emergencies

For the readers of Corriere della Sera and for all those who took part in the fourth edition of Il Tempo della Salute, Spot is not a stranger. They have already seen him at work on the stage of the event, on 12 November 2022. At the University of Genoa they started working on Spot as part of Dioniso, a project whose main objective is the development of innovative systems for the management of emergencies and rescuesthrough different types of technologies. For example, wearable devices, which can help rescuers create a map and obtain various information on the environment, or robotic systems, such as drones or the robot dog Spot.

The USARBOT project: robotics and AI for major environmental disasters

The next phase has begun in Foligno. What is it about? The USARBOT project aims to explore robotic and artificial intelligence (AI) technologies for support search and rescue operations in the face of major environmental disastersfor example of the seismic type – explains Professor Sgorbissa, coordinator of the project together with professor Carmine Recchiutowhich makes use of Zoe BettaMore, PhD student at the University of Genoa, as developer of the system. Let’s imagine, for example, the scenario in which a serious seismic event occurs in a populated area. The project explores the possibility of structures for public use such as a hospital or a shopping centre host, for preventive purposes, a quadrupedal robot equipped with the ability to carry out research tasks independently on rough terrain.

The scenarios: the robodog woken up by a seismic shock

What could happen? If the event occurs, e before the arrival of the rescuers, the quadrupedal robot will be awakened by the seismic shock, its AI will detect the emergency, and get the robot out of its recharging station to carry out tasks aimed at reducing the loss of human lives -says Sgrobissa -. To this end, the project envisages that the robot implements different behaviors depending on whether it is in the pre- or post-event phase.

In the pre-event phase, the robot will “wake up” with a pre-set frequency (for example, once a week) to carry out a night tour of the structure in complete autonomy, producing detailed 3D maps of the structure and contents. To this end, the robot uses cameras and laser sensors to construct maps. In the post-event research phasewhen the robot is “woken up” by a seismic event, these maps, coupled with sensor data, will allow its AI to autonomously explore the building looking for damage and people in distressevaluate the number of people in need of help, their position and state of consciousness (whether the person is standing or lying down; if his eyes are closed or open; if he moves or not; if he answers questions).

A software to actively search for survivors

The project also involves the development of a particular program, as Professor Sgrobissa recounts: it is important to underline that the AI ​​software that controls the robot in the post-event phase it will be developed to actively search for signs of people via not only traditional cameras but also thermal cameras and microphones. The use of other sensors among those usually used by rescuers to search for people will be evaluated on the basis of their ease of integration into the system architecture and the expected benefits. The robot will plan its route considering both the maps previously acquired in the pre-event phase and the current situation, with particular attention to the sources of risk (e.g. gas pipes, deposits of flammable material) which have been reported in advance by experts humans and are part of his prior knowledge of the environment.

Rescuers arrive at the scene

What will happen when the rescuers have finally reached the area and can directly intervene on the spot? When the rescuers have reached the mall and have to make decisions (which routes inside the partially collapsed building are safe and which are not, which sources of danger require attention, whether there are people trapped inside and how to reach them), the robot will communicate the acquired information to them, reducing the risks to which rescuers are exposed and maximizing the efficiency of rescue operations. At this stage it is also possible for human operators to take direct control of the robot, using immersive reality techniques which allow operators to “see the world through the eyes of the robot”, allowing for more effective interaction with the environment and with people in need of help. For example the robot will be able to put any injured people in communication with the health workers outside of the building to understand how to intervene in the best possible way.

So far the design aspect, but what are the results in the field? The test was also attended by personnel and volunteers from the Civil Protection and Fire Brigade of Perugia. went well – Professor Sgorbissa says satisfied -. Spot, who has always been guided at a distance, climbed ladders, entered dark rooms and tunnels, and passed over the rubble without difficulty. We acquired two types of information: how the robotic dog moves on “real” types of terrain and different from the laboratory. And we found, for example, that Spot has difficulty moving on grass. For an evaluation of the various “tasks” performed, we distributed a questionnaire to the volunteers. Furthermore, we have captured data of all kinds, leveraging the cameras, the on-board computer and the Lidar 3D laser scanner system with which our Spot model is equipped. Through the analysis of how he walks, for example, we would like to create a deep learning algorithm which will then allow us to understand on which ground he moves. the beginning of a new experience.