Ultra-fast Acoustic Detection System for transport infrastructure
Our specialists have been dealing with the latest security industrial technologies for many years, both in connection with their previous work in multinational corporations and within local Czech R&D companies.
One of our areas of interest is safety detection systems for transport infrastructure, specifically technology for road and highway tunnels. Most of the world’s technology companies have taken the path of automated analysis of images from traffic cameras. These technologies use artificial intelligence systems that are capable of automatically identifying dangerous situations using sophisticated mathematical algorithms and existing information (i.e., camera images and data from sensors in the tunnel). Unfortunately, these solutions do not immediately detect dangerous situations in tunnels; instead, there is a measurable delay. The fastest current technologies are induction loops, but these alone provide very limited amounts of information. The information delay for these technologies ranges from a few tens of seconds to tens of minutes, and during this time, hundreds of vehicles can enter the tunnel with potentially fatal consequences. Moreover, in the low-light, high-dust environments of tunnels, automated video detection systems using cameras may be prone to false alarms. Experiences from real operations show that current technologies cannot automatically identify all dangerous situations; that is, some are not detected at all. In these situations, the safety of the tunnel depends solely on the attention of the operator.
Our company complements existing systems with a unique technology that moves the resulting safety of the traffic tunnel to a higher level. Namely robotic sound analysis that uses intensive mathematical modeling, complex signal analysis, and software with progressive algorithms. Cooperation with a renowned foreign research and development partner, who works, among other things, for the NASA agency, has developed from its original business base to the level of a joint technology project, in which our company, Powerdynax, develops one complex part of this system.
For artificial intelligence systems, hardware development costs (mainly sensors) represent only a fraction compared to the development and evaluation costs of the software algorithms. The Acoustic Detection System (ADS) from Powerdynax fits precisely into this category of modern products. ADS can be operated independently, but it brings incredible additional value when combined with other safety systems to create an integrated broad-spectrum (audio and video) tunnel safety system.
The Acoustic Detection System analyzes the sound background of the tunnel pipe in real time. Small microphone boxes (MBs) are added to the locations of existing video system cameras. These exact locations of MBs are not critical, meaning they can be installed anywhere near the existing cameras since the influence of the camera on the acoustical data coming from the tunnel is minimal. The acoustic signals from the tunnel pipe are converted into digital signals, and the data are transmitted in real-time via the tunnel network infrastructure to an analysis server or a server farm. Just as the tunnel tube behaves from the acoustic viewpoint, our robotic system behaves as a whole in the resulting behavior, regardless of how many microphone boxes are installed.
In the analysis server, intensive signal analysis of all microphone channels takes place in real-time. How does this process work? Let’s use a simplified example. Please put yourself in a situation where you have the TV on and you are watching a very interesting program. In the next room, behind a closed door, your six-month-old baby starts crying. Even though the sound is faint and almost inaudible, your brain immediately reacts and assigns the sound as a potential “danger signal” and assigns it a very high priority using the dynamic filters in your brain. The brain can evaluate this information correctly, quickly, and subconsciously, and then your consciousness immediately focuses on managing any possible danger. Our Acoustic Detection System works very similarly.
If a dangerous situation is identified, the Acoustic Detection System automatically assigns the situation to predefined categories and calculates the location of the incident based on time analysis of signals from multiple microphone channels. The communication server then sends the relevant information to the tunnel control system. Based on priorities, the relevant cameras display the scene of the incident on monitors in the control room. The operator is automatically informed of the dangerous situation, and within a few seconds of the incident occurrence, the incident location is displayed, and pre-programmed follow-up actions are activated.
The Acoustic Detection System also enables automatic dynamic localization of moving people inside the tunnel using Heat Maps, which are displayed using an integrated graphical interface.
Currently, approximately 150 km of tunnels are protected by this system, and operational experience shows that it is clearly the fastest method of detecting dangers within the given categories and, at the same time, has the smallest number of false alarms. In some cases, the reaction time of the Acoustic Detection System is less than 300 milliseconds after the occurrence of the event. If you are interested in the Acoustic Detection System, please contact us by phone +420 778 468 087 or by email at firstname.lastname@example.org