TUSUR Develops Compact Vision System That Lets Robots See Through Fog and Smoke
Researchers at Tomsk State University of Control Systems and Radioelectronics (TUSUR) have developed a compact, active-pulse television measuring system (AP TMS) for navigating mobile robots and autonomous vehicles. Created as part of an independently initiated project under the Prioritet 2030 (Priority 2030) program, the device is designed to help robots and autonomous systems navigate in low-visibility conditions.

In November 2026, TUSUR presented a compact active-pulse system capable of generating depth maps in conditions of poor visibility. The device is the result of several years of painstaking work. What does the technology involve, and what challenges do engineers still need to overcome?
The Physics of Seeing Through Fog
A conventional video camera becomes effectively blind in dense fog, smoke or twilight. Light scatters off airborne particles, turning the image into a milky-white blur. Researchers in Tomsk have proposed a different approach: active-pulse imaging. A short laser pulse illuminates the scene, and a receiver records the reflected signal after a precisely controlled time delay. This synchronization filters out stray light scattered by particles close to the lens, making it possible to estimate the distance to each point in the scene.
Unlike a conventional video camera, the system generates a depth map – an image in which each area of the scene is assigned an estimated distance to the observed surface. Mobile robots need this information to identify open space, detect obstacles and plan their routes. According to the developers, the system currently produces 25 depth maps per second. After software optimization and a reduction in image resolution, the researchers expect to reach 50 maps per second or more.
The research behind the method has been underway for years. In 2023, TUSUR scientists published studies on digital filtering of depth maps. In 2024, they published a quantitative assessment of the accuracy of multizone measurements in the journal Avtometria. In July 2025, they reported a patented method for suppressing backscatter and the first tests conducted in an aerosol chamber at the V.E. Zuev Institute of Atmospheric Optics of the Siberian Branch of the Russian Academy of Sciences. The latest achievement is the miniaturization and engineering implementation of a principle that had already been validated: the system now has a compact enclosure, an HDMI output, a Wi-Fi module and a mobile app for configuration.

Russia’s Research and the Global Landscape
TUSUR is not the only institution working on this problem. In 2022, Roselectronics, part of the Rostec State Corporation, tested a hybrid video camera designed for observation in challenging weather conditions. In 2026, researchers at the Murom Institute of Vladimir State University unveiled a radio-frequency imaging system that can detect people through smoke. In the spring of the same year, an international research team that included the Moscow Institute of Physics and Technology (MIPT) demonstrated Un-ViTA Stereo, a neural network-based stereo vision method that estimates depth from images captured by two conventional cameras.
Meanwhile, researchers elsewhere are pursuing radar-based approaches. In late 2024, the University of Pennsylvania introduced PanoRadar, a system that uses radio waves and machine learning to create three-dimensional images. This points to an important conclusion: there is no universal, all-seeing sensor. Different physical approaches – active optics, radar, thermal imaging and passive stereo vision – are more likely to complement one another than to replace one another.

From the Lab to the Market
The system developed at TUSUR is not a ready-to-use component for an autonomous vehicle, nor will it be commercially available tomorrow. In the longer term, however, it could help search-and-rescue robots navigate smoke-filled buildings, industrial autonomous platforms operate in dusty factories and outdoors in fog, and driver-assistance systems gain a backup perception channel when visibility deteriorates.
Over the next few years, the hardware is likely to undergo further refinement, field testing will expand, and neural network-based object recognition will be integrated. The most realistic path forward is not to replace existing sensors but to supplement them: an active-pulse camera working alongside a thermal imager, radar or inertial navigation system. Combining these physical approaches is what could allow robots to operate reliably in conditions where any single sensor would fail.
TUSUR's development is a notable example of how years of fundamental research can gradually yield a practical engineering product. But the journey from the lab to an actual robot or autonomous loader is still ahead. The system's performance will need to be validated under the demanding conditions of a Russian winter, an industrial site and a smoke-filled corridor.









































