The project aims to develop a system that detects water loss using sound signal processing algorithms via a single sensor placed in a commonly used sink area and enables prompt intervention by a responsible staff member.  The developed water sound recognition algorithm was tested in different sinks; the experiments demonstrated that it achieved an accuracy rate of 77.3% and could prevent 15.25 liters of water waste in 135 minutes in a single restroom. Additionally, we trained an AI model to classify the sound, and our AI algorithm achieved a classification accuracy of 95%. Using a contactless approach, this system provides a low-cost, fast-deploying, and actionable solution that ensures water conservation and sustainability.

This research explores using natural sulphurous waters for various chemical applications. It focuses on precipitating heavy metal ions, eliminating free radicals, and recovering sulphur.

Key findings include successful H₂S detection, and sulphur recovery through oxidation with hydrogen peroxide. H₂S effectively precipitated several heavy metals (except chromium) and neutralized free radicals. The study demonstrates that sulphurous water is valuable for chemical reductions, metal precipitation, organic synthesis, and radical scavenging, suggesting it could be a sustainable source for sulphur recovery, reducing reliance on less healthy extraction methods.

In view of the increasing number of heavy rainfall events, we have developed WarnMe – a flood warning system. It records water levels in real time using infrared and ultrasound sensors and measures flow velocity via radar. The data is sent to our server via LoRaWAN, where it is analyzed using intelligent algorithms. A user-friendly app ensures that users are immediately informed of potential hazards. Simple installation instructions enable independent setup without professional assistance, creating a decentralized, cost-effective, and scalable system. In summer 2024, WarnMe successfully detected an imminent flood event and contributed to public safety by issuing timely warnings to the affected population.

A possible solution to the increasingly urgent protein deficiency could be green biorefining, which is a process to get edible proteins out of plants. But one of the produced by-products called brown juice is harmful to the environment in large quantities, for example, it can cause eutrophication. In my research we wanted to get advantage of this situation by using the brown juice as an additive in the growing media of microalgae to increase the production and the nutritional value of the harvested biomass. Recycling brown juice this way helps to fight with protein deficiency because the harvested biomass contains high amount of protein, suitable for human or animal consumption.

This research presents a next-generation AI-powered rapid water testing system designed for real-time, high-precision water quality monitoring in both industrial and domestic settings. At its core is the ESP32-C6 microcontroller, integrated with a 1.47-inch TFT display for immediate, on-site data visualization. The device utilizes five complementary analytical techniques; capacitance measurement, resistance analysis, ultraviolet (UV) exposure, infrared (IR) absorption, and Raman spectroscopy to evaluate water quality within seconds. These multi-modal sensing approaches enable the system to detect a wide spectrum of contaminants. Capacitance and resistance measurements help identify inorganic ions, dissolved salts, and microbial presence. UV and IR absorption offer rapid insights into organic pollutants, while Raman spectroscopy provides detailed molecular fingerprinting. Collectively, these techniques allow the estimation of key indicators such as Biochemical Oxygen Demand (BOD), Chemical Oxygen Demand (COD), total coliforms, and fecal coliforms. The system can also identify heavy metals, synthetic dyes, microplastics, and pathogens like E. coli in real time. An embedded AI model, trained on a large and diverse dataset of water samples, interprets the data to detect complex contamination patterns with high reliability. This results in a portable, energy-efficient, and cost-effective solution capable of delivering immediate water quality assessments, empowering users to make proactive, informed decisions regarding water safety. However, the system does have certain limitations. It cannot measure all possible water quality parameters such as specific pesticide residues with high accuracy due to the constraints of sensor selectivity and hardware resolution. Moreover, while the device performs well in a broad range of scenarios, it may require calibration or additional modules to ensure accuracy across highly specialized or extreme conditions.

Water faces serious threats, and leaks drive the need for innovative solutions. In Anabta, 40% of water is lost due to leakage (Anabta Municipality, 2021), including a 20-year-old undetected leak that caused street collapse. In the West Bank, non-revenue water reached 62% in Anata due to illegal connections, and 51–55% in other areas like Jenin and Bal’a due to leaks. In Gaza, losses reached 58% in Al-Maghraqa. Globally, Bangalore and Tokyo suffer significant leak-related losses.

Our robot, EDM, uses Arduino WeMos D1 with sensors along pipelines and a Raspberry Pi mobile robot with a camera and servo arm to detect and remove invasive roots. Data is sent via network to authorities, and leaks trigger automatic pump shutdown—advancing SDG 6 and SDG 11 in Palestine.

Oil spills devastate marine ecosystems, and conventional cleanup methods are often dangerous and inefficient. Our project, the Oil Recovery and Cleanup Autonomous (ORCA) system, presents a safer, smarter solution. This fully autonomous prototype integrates AI for oil detection (97% precision), LiDAR-based navigation for obstacle avoidance, and an optimized oleophilic surface for collection. Rigorous testing confirmed high skimming efficiencies up to 75.7% and excellent water rejection of 96%. ORCA demonstrates a viable, next-generation approach to oil spill remediation, capable of operating effectively without risking human lives, protecting both our oceans and the people who work to save them.

Many cities suffer the consequences of flooding after a large amount of water
falls in a short period of time. Therefore, this project determines how urban
vegetation modifies soil structure, improving its water infiltration capacity and
reducing flood damage, putting Nature-Based Solutions (NBS) into practice.

A humidity sensor was built using Arduino and the humidity and water
infiltration capacity of soils influenced by five tree species and seven different shrub
species were measured. Analysis of the results obtained shows that areas
occupied by shrubs infiltrate a greater amount of water. In this sense, species such
as rosemary (Salvia rosmarinus) and oleander (Nerium oleander) stand out for the
development of their root system.

With this information, an image recognition model was created and trained
using LearningML, capable of identifying the studied plants in situ and indicating the
infiltration capacity of the surrounding soil. From it, a Geographic Information
System (GIS) was generated using the QGIS program, which shows the areas of our
city most vulnerable to torrential rains and flooding.

My project focused on detecting heavy metal pollution in rivers using fish scales as an environmentally friendly sampling material. I collected chub scales from sites along the Jihlava and Oslava Rivers, both upstream and downstream of wastewater and mine water treatment plants. The scales were cleaned, digested, and analyzed using ICP-MS. I measured concentrations of lead, copper, and cadmium, and assessed how they vary with fish size and sampling location. The findings suggest that treatment plants may contribute to pollution. Fish scales proved to be a sensitive and non-lethal tool for monitoring heavy metal contamination in aquatic ecosystems.

In this scientific research work, the hypothesis was tested, whether it is possible
to obtain drinking water for human consumption by filtering the freshwater through
natural materials. As part of the research, experiments were carried out in which the
water sample was filtered through sand, charcoal, and moss. After filtration, the water
was boiled at 100 °C for 10 minutes. After boiling, the water samples were tested in the
Scientific Institute of Food Safety, Animal Health and Environment BIOR (Latvia, Riga). By
comparing the obtained test results, it was concluded that these filter materials,
combined with boiling, have great potential to filter drinking water for short-term use.

Waver is an innovative aquatic drone designed to restore and monitor dead zones in aquatic environments by generating artificial waves and collecting real-time environmental data. Equipped with smart sensors, solar panels, and GPS, Waver operates autonomously while monitoring key parameters such as temperature, pH, and oxygen levels. It integrates three AI models specialized in detecting plastic waste, oil leaks, and marine life, supported by a robust dataset for high accuracy. A user-friendly online platform displays live data and alerts, enabling informed decision-making for environmental management. Waver aims to support biodiversity recovery, promote sustainability, and offer a scalable solution to one of the most pressing ecological issues aquatic ecosystem degradation.