Artificial intelligence surprised scientists with its development.

Researchers at the OpenAI laboratory in San Francisco conducted an experiment testing two ideas: a distributed learning system consisting of competing or collaborating algorithms to test the emerging behaviors, and the "reinforcement learning", i.e. learning how to achieve a goal through trial and error (each success is rewarded, each failure is punished). In the experiment, so-called autonomous agents, i.e. instances of AI algorithms, were playing hide-and-seek game. The researchers wanted to observe the evolution of decision-making mechanisms. The hiding and seeking teams of the autonomous agents could operate in a virtual environment consisting of a closed space with various objects, such as walls, ramps and boxes. Each item could be blocked by the hiding team that, just like the classic hide-and-seek game, had a few seconds of advantage over their rivals. The researchers did not program the agents in any other way - their task was to develop appropriate strategies. According to the reinforcement learning technique, the steps leading to the goal, i.e. finding the hiding ones, were rewarded, the steps leading to losing were punishable.

Over the course of hundreds of millions of rounds of the game play, the agents developed several strategies and counter-strategies. During the first phase, the hiders and seekers developed very crude avoiding and chasing strategies by moving around the space without manipulating any of the objects. But after the 25-million-game mark, the hiders learned to move and lock the boxes and barricades in the environment to build forts around themselves so the seekers would never see them. After another 75 million rounds, the seekers discovered a counter-strategy. They learned to move a ramp next to the hiders’ fort and use it to climb over the walls. But 10 million rounds later, the hiders had learned to lock the ramps in place before building their fort.

At first, the researchers at OpenAI believed that this was the last phase of the game play, but finally, at the 380-million-game mark, two more strategies emerged. The seekers once again developed a strategy to break into the hiders’ fort by using a locked ramp to climb onto an unlocked box, then go their way on top of the box to the fort and over its walls. In the final phase, the hiders once again learned to lock all the ramps and boxes in place before building their fort.

The experiment proved that the combination of competition and learning principles with reinforcement works in the digital world. The developed solution surprised the creators and is a positive signal for further attempts in more complex environments. Many artificial intelligence researchers believe that reinforced learning can be used to solve complex tasks in the real world.

Winners of the 18th DIPOL photo contest.

We thank all participants for the submitted photos. We received a lot of amazing pictures, which made the task of awarding prizes in this competition a particularly difficult job. Congratulations to the winners and honored participants – you have proved that such inconspicuous and seemingly common objects as antennas can be inspirations for creating truly artistic works.

1st place
Dochodzi piąta – Adam Sowa
[The fifth]

Control of a driveway gate with IP Villa video intercom.

Villa gate/door stations have relay output for control of a door/gate electric strike. To be able to control an additional car gate, it is necessary to install additional door control unit – Hikvision DS-K2M060 G74818. The unit has to be connected with the door station via the RS-485 interface. The relay output of the module (powered with a 12 VDC source) should be connected to the control input of an automatic driveway gate. The system should be configured with the iVMS-4200 client application.

An example of IP Villa video intercom installation using additional module for the control of the driveway gate

Hybrid multiswitch system.

Most satellite receivers available in the market support SCR/Unicable technology, supporting EN50494 and/or EN50607 standard(s). Along with a lower number of cables, a big advantage of this type of installation is the flexibility in terms of system topology that can be similar to signal distribution systems used in terrestrial TV systems. There is no requirement for star-like typologies and existing terrestrial antenna systems can be easily adapted to receive and distribute satellite signals (of course, taking into account the maximum number of users connected to one multiswitch).

Signal Fire fusion splicers – DIPOL extends the warranty to 4 years.

The growing popularity of optical fibers in building installations and access networks forces installation companies to expand their business profile to include optical connections. The basic professional tools used to implement such installations are fiber optic fusion splicers. DIPOL company offers two models of fusion splicers, Signal Fire AI-8C L5870 and Signal Fire AI-9 L5875. These devices are perfect for both the beginner and experienced installer. Due to a simple operation and low price in relation to their functionality and technical excellence, Signal Fire splicers are also used in many technical colleges and universities for instructing students how to correctly splice optical fibers.

Each installer buying a Signal Fire splicer has the opportunity to undergo a short training in Krakow on the principles of operation, use and maintenance of the splicer.

We offer first 5G antenna!

TRANS-DATA 5G 7/7/7 A741025 is a directional panel antenna intended for outdoor use, designed mainly for LTE/5G modems. Depending on the band, the gain of the antenna ranges from 6.5 dBi to 7 dBi. The antenna has been designed for operation with LTE/5G modems equipped with two antenna connectors, some cellular phones, GSM, DCS, 3G modems.

TRANS-DATA is a brand of high quality antennas for wireless reception and transmission of digital data. Thanks to high quality materials and very good performance, the antennas are characterized by reliable and stable parameters. TRANS-DATA antennas significantly improve signal strength and increase the coverage area.

Key features of the TRANS-DATA 5G 7/7/7 A741025 antenna:

• Panel antenna for outdoor use
• MIMO 2x2
• Broadband: 698–960 MHz, 1710–2700 MHz, 3300-3800 MHz
• Multipurpose: GSM, DCS, UMTS, WLAN, LTE, 5G
• Gain:
• 7 dBi (698–960 MHz),
• 7 dBi (1710–2700 MHz)
• 6.5 dBi (3300–3800 MHz)
• Shorted for DC
• Low VSWR: <1.8
• 30 cm lead with N socket
• UV resistant and weatherproof ABS plastic cover

TRANS-DATA 5G 7/7/7 A741025 antenna

ImagePerfect firmware of Seagate SkyHawk hard drives for high image integrity.

By using "ordinary" hard drives in a video surveillance system, you may risk losing data. The problem results from the fact that hard drives used for video monitoring work in different conditions than the drives used in PCs. In CCTV systems, the recording of data takes place on a continuous basis and accounts for over 90% of the disk load. The great challenge is also the need to simultaneously handle many data streams generated by individual cameras.

The algorithm of collecting data in the cache and writing to disk after buffer filling

Seagate SkyHawk hard drives work under the control of special firmware for CCTV drives. It is based on multi-tier caching technology (MTC) ensuring high image integrity and simultaneous recording of up to 64 camera streams. The ATA-8 protocol is optimized for recording sequences of large data blocks. Data from the DVR/NVR are stored in the disk cache memory. The writing operation to the non-volatile hard drive’s media takes place only after the buffer memory is full. This results in fewer number of data write/read operations, which translates into greater reliability.

Large cache memory allows for limiting the number of write operations

The efficiency of large file transfer is not only associated with high bandwidth requirements, but also the availability of robust error correction functions to ensure video data consistency during transmission. The discs use a special data correction algorithm operating on the fly, limiting the possibility of losing image frames during recording. The disk software cooperates with the DVR/NVR in such a way that it does not hold up any requests from the host system, ensuring appropriate query handling time. This minimizes the risk of losing frames during recording and of freezing issues during playback.

NVR-based integration of intelligent functions of IP cameras and PTZ camera. Hikvision IP cameras even from budget series (except EasyIP Lite) allow for intelligent image analysis in the form of detection of virtual line crossing and area intrusion. The functions can be activated and configured after logging in to the camera with the use of the iVMS-4200 client software or directly from a Hikviosion NVR. The virtual line crossing and area intrusion detection can be used to detect people, vehicles and other objects. In response to the crossing of the line in a defined direction or intrusion into a defined area, the system can inform about the event, trigger video recording or alarm output, send an email, etc. These events can also be linked with PTZ camera in order to trigger a preset, patrol or route... >>>more

IP video surveillance system using intelligent functions of compact cameras and special features of a PTZ camera. After an intrusion into the area monitored by a fixed camera, the PTZ camera goes to the predefined preset, showing the details of the event.

Vertical mounting system in 19" RACK cabinets