The European Consortium for Mathematics in Industry (ECMI) and Department of Mathematics and Informatics Univeristy of Novi Sad invite teams of students of Bachelor/Engineering, Master or PhD levels in an ECMI member institution to the ECMI 2026 Student Competition. Each team should present its solution to a mathematical problem of social or economic importance.
Who can apply?
Teams of 1 to 4 students of Bachelor/Engineering, Master or PhD levels in at least one ECMI member institution during the school year 2026/2027 can participate. The list of ECMI members is available at https://ecmiindmath.org/list-of-ecmi-members/. Each team chooses a name for identification. The team should nominate a leader who will present the results, but registration of all team members is mandatory.
Prize
The winning team will receive 1000 EUR cash prize, sponsored by ECMI, and will be invited to give a presentation during the 40th ECMI Modelling Week, Wuppertal, Germany.
Problem: Modelling and prediction of anomalies in autonomous vehicles
An autonomous vehicle is driving through the city of Novi Sad, while continuously collecting GNSS and IMU data. Under normal conditions, the vehicle follows a predictable route; however, while driving, one of the sensors begins to malfunction, causing sudden irregular acceleration values and unusual magnetic field measurements that do not match the vehicle’s actual motion. Although the vehicle continues moving, these subtle deviations may indicate an anomaly that could lead to unsafe behaviour if ignored. This unfortunate turn of events proceeds to occur while driving the car relatively often, which makes the usage of this convenient car impossible on the roads of Novi Sad.
Foreword
AI-driven autonomous vehicles are widely regarded as a key component of the future of transportation. Autonomous vehicles operate in dynamic environments where unexpected situations can arise at any moment. Even small deviations from normal behavior – such as unusual sensor readings, abnormal motion patterns, or subtle system malfunctions—can quickly escalate into dangerous scenarios if left undetected. Detecting anomalies early allows intelligent systems to respond appropriately, whether by alerting operators, switching to a safe mode, or taking corrective action to avoid accidents.
Data
The provided dataset was generated using NB-IOT edge nodes mounted inside an autonomous road vehicle. Each edge node collected data locally while the vehicle moved through city. The vehicle maintained continuous connectivity to the cellular network, allowing synchronized time stamping and uninterrupted data logging throughout each route. Positioning information was obtained from a Global Navigation Satellite System (GNSS) module, including timestamp, latitude, longitude, altitude, vehicle speed, and the number of satellites in view. In addition, data from the onboard Inertial Measurement Unit (IMU) were recorded, capturing acceleration and magnetic field measurements along all three spatial axes. GNSS data were sampled at a temporal resolution of approximately 10 seconds, while IMU measurements were collected at a higher frequency of approximately 15 milliseconds. For each GNSS sampling interval, the IMU data were aggregated by computing the arithmetic mean and root mean square (RMS) values for both acceleration and magnetic field signals. The data is given as a time series table: training.csv, test.csv, and additionally, a file anomalies.timestamps is provided, which includes the test data ground truth anomaly timestamps for model evaluation. The data is available here.
Task
The task is to derive an anomaly detection model for autonomous vehicles using whatever approach is deemed more effective by the participants. This might include analytic, semi-analytic approaches, deep learning, statistics, modelling, and numerics.
Registration
- Teams of 1 to 4 students of Bachelor/Engineering, Master or PhD levels in at least one ECMI member institution during the school year 2026/2027 can participate. Each team chooses a name for identification. The team should nominate a leader who will present the results, but registration of all team members is mandatory.
- The list of ECMI members is available at https://ecmiindmath.org/list-of-ecmi-members/.
- The registration form is available here and should be sent to student.competition@ecmi-indmath.org.
Deliverable
- A written report (in English and PDF format) of at most 12 pages is the expected result. The report should describe the analysis strategy, the methodology used, and the interpretation of the results, highlighting the main findings.
- In case computer simulations are performed, they should be well documented with results included in the report and accompanied by the code. The code should be written in Matlab/Octave, Julia, C, Python or R.
Criteria and deadlines
Criteria
The Competition Committee of the ECMI Student Competition will judge the originality of the approach, the quality of the implementation, the quality of the results, and the overall quality of the report. The Competition Committee reserves the right to ask shortlisted teams to present their report findings and answer questions about their solution. The use of AI tools requires explicit disclosure and explanation.
Deadlines
Problem posted: June 27, 2026
Registration due: October 15, 2026
Submission due: February 28, 2027
Winners announcement: May 15, 2027
ECMI Student Competition Committee
- Prof. Paola Causin, Dipartimento di Matematica, Università degli Studi di Milano
- Prof. Marta Pascoal, Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano
- Dr. Marek Teuerle, Wroclaw University of Science and Technology
- Luka Rutešić, University of Novi Sad