Robodog

A robodog or an automated system of industrial inspections and monitoring “Cross Cyber-dog” is designed to solve security problems on the factory floors, automation of monitoring of industrial inspections and in other areas that require remote control and monitoring of the perimeter.

Pressure gage readings using computer vision system

The ability to implement readings from various devices on condition that the object can be captured at 45˚ angle.

Leakage identification.

The ability to diagnose leakages of chemicals at a monitoring site - visual, through CV, and considering the readings from an environmental sensor.

Anomalous sound detection.

The ability to use computerized learning of sound indicators from highly sensitive microphones of the surrounding perimeter - car engine, launching a drone and unmanned aerial vehicle, speech recognition (Yandex.Speech Kit - Internet is required) in the sensitivity mode of audio equipment.

SF6 gas detection

Gas leakage detection.

Radiation detection.

Ability of changing the radioactive background of the monitoring site.

Thermal control.

Control of the heat map of a given status of the monitoring site, presence of facilities variating the temperature condition on the perimeter relatively environmental condition.

Crack detection

Monitoring the integrity of the structures of the monitoring sites - buildings, walls, rooms, steel structures, monitoring changes vs the reference conditions of the structural health.

Face ID feasibility

recognition of friend / foe, vehicle license plates, based on CV algorithms. Based on the customer’s requirements, it is possible to train neural networks based on the date set according to the preset parameters.

Perimeter monitoring independence:

Use of Deep learning

algorithms to control variation in the relevant perimeter from the reference model with the ability to bypass the specified route.

Navigation planning based on the route map

variation of the robot’s route in real time from the control unit of the dispatch service.

OctoMap (probabilistic map).

An efficient probabilistic 3D mapping framework based on octrees for detection and recognition of environmental medium (quadcopter, controlled car and gearbox).

Dynamic obstacle perception

The robot avoids obstacles and carries out probabilistic prediction of the perimeter control route reconfiguration based on predictive models using auxiliary systems.

Creating 3D maps in real-time

To fill up the decision-making system with additional information and integration with related systems (for example, comparing the metrics of the thermal variation in the environment vs the given characteristics, remote analysis of indicators at the control unit). Control of presence of objects, control of variations in the perimeter of the guarded object and registration of foreign objects when detecting the objects in the reference 3D model of the perimeter (Lidar technology - boxes, movement of flight hardware and wheeled vehicles).

Global / local positioning

additional control while device movement positioning afield. Recording the efforts of electromagnetic action during GPS Spoofing attacks, subject to comparison of the identification data of OctoMap objects state.