albfaro@dscomp.cloud

© dscomp2019

• Home Page

dscomp.net

DEEP SENSING COMPANY

Ubiquitous and Pervasive Sensor Nets even powered by AI and Deep Learning 

OUR SCOPE

Develop and commercialize pervasive and ubiquitous sensing products and infrastructures based on Cyber Physical Systems (CSPs), even powered by AI technologies including Deep Learning, to support demands of intelligent quasi-autonomous control coming from emergent green and blue markets, e.g., home solar energy management, elderly care and people security, DOP food in agriculture, and sustainable mobility.  These sensing platforms are conceived to be included in the CPS model to guarantee technical, semantic and user interoperability.

CPS Model (A.Faro&alii, CES LasVegas 2016)

DEEP SENSING TECHNOLOGIES

Our sensing platforms for system control are developed by using specific sensing nodes, either new or available on the market, able to send the sensed data at distance through internet. Sensed data range from simple ON-OFF values to complex patterns such as images taken by camera. The devices are governed by rules.  In the complex cases they learn the suitable behavior from data. Rules  and learning algorithms are embedded into the leaf sensing nodes for responding in real time to the monitored events or at intermediate or final coordination nodes in case the time is a less critical condition. Low cost technologies are used   to satisfy the typical consumer demands, e.g. the ones coming from the domotics and automotive markets. Advanced sensing systems embedding deep learning are produced for specialized projects, e.g. the ones involving image processing. Our ultimate goal is to develop and commercialize these deep sensing devices for consumer applications too.

Basic sensing technologies including Image processing.

Healthcare and Telecare

Contact and IR Sensors

Bio-data are sensed by either available BLE sensors on the market or by DIY edge devices. They deal with Body temperature, Blood pressure, Oxygen in the Blood, Dehydration, Heart Rate Variability (HRV), Mental Stress, Breathing Rate. Such sensors are able to process data to carry out the first real time control and are provided with internal gateway to send data to brokers resident on remote control centers. Currently such intelligent sensors are tested in the Projects IHOSP and SELCA supported by EC Regional Funds.

Pervasive & Ubiquitous Control

Control by Images and Video 

Hospitals and Clinical Laboratories make available  the devices that are needed by remote patients, and vice-versa patients make available to a remote hospital the home devices or the ones they wear while are walking or driving  to monitor/control their health status or pathology from distance.   Currently this monitoring and control system is tested in the Projects IHOSP and SELCA supported by EC Regional Funds.

Thermal and optical IP cameras  are used  to control the facial temperature and the posture of a people. Also people  identities are checked when they  pass closer to the camera. Images and video are stored on SD or remote servers for statistical processing.  Currently these cameras are tested in the Project IHOSP and BioTrak supported by EC Regional Funds.

Alternative Energy

Such low cost smart system aims at controlling that the energy produced by the domestic solar plant is completely exploited by the domestic appliances, otherwise the exceeding energy is used to charge some service batteries exploited in the evening or during the night to decrease the electrical consumption of the house. Two main sensing devices measure in real time the electrical consumption of the house and the energy produced by the solar plant. The energy coming from the solar plant is consumed by the house appliances, but when it exceeds the consumption demand, the exceeding collected energy is used to charge progressively some service batteries connected to the home electrical circuit. The batteries can be located in every part of the house.    The plant status can be monitored and controlled via internet from everywhere by using a simple APP . The system is sold preferably to the electrical installers.

Battery charge from everywhere in the house

Home solar plant control scheme

Domotics

Our platform for Domotics  deals mainly with the surveillance of the typical aspects related to home security, e.g, the detection of unauthorized people, water leakage, bad weather conditions, sudden disconnection of the power supply, low voltage of the service batteries. All these aspects are monitored by low cost devices that can be  integrated with relevant technologies available on the market, e.g. Arlo and Reolink Cameras, Samsung  Smatthings,  and Sonoff Bridges. As much as possible of integration will be accomplished case by case by using suitable technologies (e.g., MQTT, IFTT, etc.).  Alerts are sent to the users for every suspicious event detected by the sensing devices. Control actions are taken immediately, e.g., the water supply from tanks or from the mains is switched off in case of water leakage, the siren is turned on in case of intrusion.  The home status can be monitored and controlled via internet by using a simple APP. The system is sold to either the end user or to electrical installers depending on the complexity of the desired control.

All the relevant technologies for domotics may be integrated in our platform

Automotive: Cars and Boats

Automotive applications deal with surveillance systems from remote to improve people security on both cars and boats. Some domotics control is implemented also in the automative field e.g, the detection of unauthorized people, water leakage, bad weather conditions, sudden disconnection of the power supply, low voltage of the service batteries. Further specific controls deal with  monitoring the collisions between adjacent moored boats or the  low pressure of the car tires. All these aspects are monitored by simple low cost devices that make available the relevant information to the users through internet APP. Alerts are sent autonomously to the users and control actions are taken immediately, e.g., the water supply from tanks or from the mains is switched off, further safety fenders between moored boats are automatically inflated in  case of bad weather conditions, battery charge is powered by the dock of the port in case of cloudy weather. The system is sold to the end users or to electrical installers depending on the control complexity.

Mobility and Logistics

1. Drivers should be informed about the current shortest paths to destination and alternative to avoid being trapped in the traffic jams signaled by cyber-physical-social systems. A set of algorithms has been developed to solve timely the All Pair Shortest Paths (APSP) problem in both the free flow and congested traffic regimes, for road networks of medium-large size, thus enabling Location Based Systems (LBSs) to deal with emergencies and critical traffic conditions for city networks ranging from some hundreds to many thousands of nodes. A parallel version of the algorithms is developing using GPU processing infrastructures to solve the APSP problem for very large areas.
2. Also, a suitable architecture has been developed for traffic monitoring in real time. It is based on a high-speed communication network whose nodes are equipped with fuzzy processors and cellular neural network (CNN) embedded systems so that the parallel APSP algorithm may provide  timely information to drivers using both traffic perceptions sent by people and traffic video-sequences taken from Webcams. 
3. Traffic information are also used to support fast and safe delivery of goods by guaranteeing that temperature and humidity conditions during the delivery process in the containers should satisfy the ones needed to maintain the quality of the goods.  Currently our ICT technologies for  controlling the delivery chain are  tested in the Project BioTrak  supported by EC Regional Funds dealing with  the delivery of high quality agri-food products .

Suggested paths to avoid to be trapped in traffic

Traffic regimes detected by cameras (A.Faro&alii, T-ITS 2012)

From Incubators and Research Labs to
Technology Transfer and High Computing Services

Deepsensing is a member of the NVIDIA Inception Program for Startups and of the Oracle Startup Program to improve the technologies  offered to their clients in the field of Artificial Intelligence and  AIoT (Artificial Intelligence of Things), and to offer intensive computing  services and "data as service" platforms to the clients. This is obtained by using intelligent and powerful edge devices based on Raspberry and  NVIDIA Jetson boards, and by means of deep learning technologies running on super computing nodes such as the NVIDIA A100 available at the Oracle Cloud.    DeepSensing Company operates to transfer deep sensing based services and products available at research labs as proof of concepts to market. Current technology transfer projects deal with elderly care from remote and knowledge discovery in bioinformatics , identifying genetic markers  for DOP foods and computer vision based sensors in agriculture. In all such projects we plan to use not only basic sensing technologies but also GPUs with the aim: a) to develop   very large and pervasive sensor nets powered by real time AI algorithms embedded in both the leaf sensing nodes and in intermediate coordination nodes, and b) to learn off line how to improve the monitored system based on the collected data, thus supporting complex diagnosis and control decisions to be injected into the sensing net. Prototypes and experimental studies are currently tested in the projects IHOSP, SELCA, and BIOTRAK supported by EC Regional Funds.

E-Commerce coming soon

Our products and services are available for companies and installers since 2020. E-Commerce for customer market is planned from 2021 at deepsensingcompany.com in cooperation with international partnerships. Starting commercial products will be in the field of ICT medical devices and IP cameras for pattern recognition using both optical and thermal intelligent sensors.