Tornyol Develops 40-Gram Autonomous Drones That Hunt Mosquitoes Mid-Air Using Doppler Ultrasound and Sound Detection
American startup Tornyol, supported by the Y Combinator accelerator, has unveiled a miniature autonomous drone designed to hunt mosquitoes in mid-air without the use of chemicals. The 40-gram aircraft is intended to locate targets by the sound of their wingbeats, calculate an interception trajectory, and physically collide with the insect. During the first successful test conducted on 14 July inside a closed facility, the prototype autonomously pursued and struck a flying moth, marking what founder Alex Toussaint called the company’s first “aerial defeat of a target.”
The demonstration still relied on external infrastructure: a motion-capture system tracked infrared markers on the drone while a table-tennis ball served as a proxy for the insect’s position. A ground computer processed the sensor data and issued flight commands. Within the next several weeks, Tornyol plans to migrate all processing to the drone’s onboard electronics. The final design will incorporate ultrasonic transducers similar to automotive parking sensors and smartphone-style microphone arrays.
Once airborne, the drone will emit ultrasonic pulses and analyze returning echoes. The characteristic frequency shift caused by rapidly beating mosquito wings—the Doppler effect—will allow onboard algorithms to detect, classify, and even determine the species and sex of the target. After identification, the aircraft will plot an obstacle-avoiding route and deliver a precise physical strike. The company ultimately envisions coordinated swarms of dozens of such drones patrolling entire urban districts.
Founders Alex Toussaint and Clovis Piedallu claim the technology could cut the cost of mosquito control by a factor of 100 compared with existing methods. According to their estimates, a formation of only ten drones would be sufficient to clear mosquitoes from one square kilometer. Although these performance figures have not yet been validated in real-world city environments, the startup continues to refine both the sensing suite and the autonomous navigation stack.
Despite the promising indoor result, significant engineering challenges remain. Engineers must still prove that the tiny platforms can reliably differentiate mosquitoes from other insects, operate safely near humans, and maintain effectiveness when flying outdoors amid wind, obstacles, and variable lighting. Tornyol’s work therefore represents an early but concrete step toward chemical-free, drone-based vector control.
Related articles
InfotecsTech Builds Custom Kubernetes-Based Traffic Generator for NGFW RnD and Performance Testing
InfotecsTech developed an in-house traffic generator to support development and testing of its high-performance NGFW cluster in active-active mode. The team rejected commercial solutions from IXIA and Xinertel due to high cost, insufficient flexibility for complex NGFW functions, and geopolitical restrictions. The resulting platform runs on Kubernetes with a master node managing Registry, Discovery, and Crux components while worker nodes host containerized generators. Supported generators include Cisco TRex for throughput and connection testing, SIPp for VoIP scenarios, pyftpdlib-based FTP generator, Yandex Tank with Nginx for live TLS traffic, and Selenium-based legitimate clients against OWASP Juice Shop. Practical scenarios cover VoIP call storms, maximum concurrent connections, 400 Gbit/s UDP throughput, 5 million CPS, and IMIX traffic at 300 Gbit/s with packet loss analysis. The system integrates Camunda for full automation of test scenarios and device configuration.
Why AI Chatbots Misread Polished Reports and How to Prepare AI-Ready Content
Beautifully designed reports often confuse AI systems because visual layout does not preserve logical relationships between elements. When design is stripped away, machines may lose connections between headings, numbers, tables, and footnotes, leading to incorrect interpretations of key facts such as revenue growth. The solution is to create AI Ready content that maintains structure, semantics, and context even after text extraction or copying. This approach aligns closely with web accessibility standards from W3C and benefits both human readers using assistive technologies and automated analysis tools. Organizations are advised to use tagged PDFs following PDF/UA and ISO 14289-2:2024, provide data in XLSX or CSV alongside visual charts, and ensure every important figure travels with its full context including period, unit, and comparison base. The same principles apply to presentations, press releases, websites, and multimedia content.
Statistical Analysis in OSINT: Tools, Methods and Real-World Intelligence Applications
The article explains why statistical processing has become an essential component of professional OSINT work when analysts face large volumes of raw data. It outlines five core tasks that statistical methods solve: actor profiling, disinformation monitoring through time-series and graph analysis, financial intelligence, geospatial verification, and threat assessment. The text compares popular tools including Maltego, Gephi, Python, R, Power BI, Tableau and SpiderFoot, stressing that real investigations usually combine several of them. Detailed examples show how Pearson correlation, Louvain clustering and TF-IDF or BERT embeddings help identify coordinated botnets in social media. Famous leaks such as Panama Papers and Pandora Papers are presented as landmark cases where regression models, cluster analysis and network graphs exposed hidden ownership structures. The piece also lists open statistical sources from national agencies, international organisations and technical platforms, and reviews key mathematical techniques from basic descriptive statistics to ARIMA, CUSUM, DBSCAN and dimensionality reduction methods.
Bot Traffic Overtakes Human Traffic in 2024 as AI Agents and Scrapers Surge
Analysis of internet traffic from 2013 to 2026 shows automated bots steadily eroding human dominance online. Imperva data reveals human traffic fell to 47 percent by 2025 while malicious bots reached 40 percent. Good bots such as search crawlers remain stable, but gray AI agents and scrapers now drive much of the growth. Companies face rising infrastructure costs from bot traffic that generates no revenue, described as an invisible tax. Cloudflare and Akamai reports confirm high volumes of automated requests, with many classified as harmful scraping. The trend raises concerns about a synthetic internet shaped more by AI recommendations than human activity.