[:ru]Отправить сообщение[:en]Send message

Intelligent systems and algorithms

Development of a car orientation and direction finding module for a mobile application

LLC “Connectivity” has developed a software module for processing data from the accelerometer, gyro, magnet compass and the smartphone GPS-receiver. The main objective of the module is to define over-a-distance vehicle-tilt angle and to identify car motion parameters (acceleration/deceleration zones on the trajectory of motion and their characteristics, lateral accelerations, turn and lane-change detection). To solve this task the data clearing algorithm was developed, as well as the algorithm estimating values of car motion parameters.

Development of a road sign information and lane departure warning system

Together with specialists of LLC “NPP SATEC plus” we implemented a real-time road sign information and lane departure warning prototype system via one video camera installed in the car.

This project is presented on our web-site as a demonstration of capabilities of our design engineers in the field of video and photo image intelligent processing.

Automatic recognition of real world objects such as road signs and lane is a difficult and science intensive task: variety of weather conditions, flecks, blur caused by fast motion, nonconformance of signs and lane marking to GOSTs etc., set very high requirements to intelligence of algorithms.

Exhibition robot "Potapych bear"

– Specialists of LLC “NPP SATEC plus” have developed a robot “Potapych bear” and the software making the stuffed toy “alive”.

– The robot is able to detect faces of the people standing in front of him at certain distance and to hold a dialogue with them analyzing movements of their heads (agreement/disagreement nods), as well as to synthesize speech, to play presentations, video and music.

– The robot is able to turn its head and to move its paws. Head turning provides the interlocutor’s face tracking in the video stream, which intensifies the effect of a dialogue with the robot-bear.

Program "Defecto-engineer"

– During development of an automated part quality control system the task came up to develop a program for the part surface defect inspection stand based on the high-resolution photos obtained an the stand.

– Specialists of LLC “NPP SATEC plus” were involved in solution of this task. Within short timeframes expandable software had to be developed for processing and analysis of work piece and machine components halftone images including not only image processing modules but also report generating modules as to the completed analyses and those of databases describing generic experience and visual defect images.

Robot articulator

– Specialists of LLC “NPP SATEC plus” developed robot kinematics, robot control MCU software, and PC software which allowed to load human jaw motion paths obtained on the third party equipment (stomatological scanner). The first trial mock-up of the device had to be implemented and upgraded in order to increase smoothness of movements and their accuracy.

Development of the program for quadrotor automatic landing on the marker

In the framework of the contest held by the CROC Company designers of LLC “Connectivity” together with engineers of LLC “NPP SATEC plus” developed software for multi-rotor UAV tracking the marker video image and computing the landing trajectory in its center.

The solution of the task was aggravated not only by absence of environmental restrictions (lighting, location of the marker), but also by the limits of the quadrotor hardware (unstable frame rate, multiplexing of video streams with non-assured time delay, low accuracy of on-board inertial navigation system, delay in actuating paths). All this required developing specific algorithms and debugging their real-time implementations.

Development of the concurrent mapping and navigation system of an unmanned aerial vehicle group

LLC “NPP SATEC plus” with contributions from engineers of LLC “Connectivity” performed research in the sphere of autonomous video stream navigation of an UAV group. The problem featured through the lack of communication with global navigation systems (UAVs were indoors), i.e. there was no possibility to use GPS or GLONASS to define the location. The location of each particular UAV was defined by it off-line (based on video stream processing) relative to the visual markers (QR-codes). Then quadrotors exchanged information to generate a unified map of indoor markers.