A new security control centre, AI-supported crowd monitoring and shorter lines of communication: following last year’s experience, Munich has further developed its security concept for the Oktoberfest. Alongside traditional threat prevention, the focus is increasingly shifting to how early critical developments can be identified and how quickly effective measures can be derived from them.

When the 191st Oktoberfest begins on 19 September, security at the Theresienwiese will be organised more than ever before via a shared situational picture. The police, the city and the organisers have responded in particular to the occasional overcrowding of part of Wirtsbudenstraße during the 2025 Oktoberfest. Monitoring management, visitor management and communication have been revised, whilst additional technology is being deployed.

However, this does not mean that the security concept relies solely on increased video surveillance. Rather, it is characterised by greater integration of camera technology, crowd monitoring, human situation monitoring, central coordination and operational personnel.

New control centre consolidates information

A key component is the new security control centre with a coordination room in the Theresienwiese Service Centre. The new area has 34 seats, 30 of which are digital workstations. Around 22 square metres of screen space allow for the simultaneous display of various information sources as well as the integration of external camera and data systems. In addition, there is a dedicated server room, areas for telephone and radio communication, and a broadcast booth for announcements.

This creates a technical and organisational platform through which information from different areas can be consolidated. This should enable the police, fire service, ambulance service, security officers, the events office and the festival management to assess developments as far as possible on the basis of a shared situational picture.

This is particularly crucial when there are large crowds. A camera image alone does not answer the question of whether intervention is required. Only by combining several sources of information and contextualising them is it possible to distinguish between a normal crowd and a situation that is taking a critical turn.

More than 50 cameras for crowd monitoring

For the city’s crowd monitoring system, more than 50 cameras are available at 35 locations across the festival grounds. Their footage can be analysed using AI. Separately, the police operate 56 of their own video cameras at the Wiesn to carry out their duties of security and law enforcement.

The crowd monitoring is scientifically supported by the Fraunhofer Institute for Optronics, System Technologies and Image Analysis (IOSB). Using the GeoVID experimental system, video footage is analysed using artificial intelligence to assess crowd densities and movement flows.

To do this, a camera view is simplified and divided into a grid. The system determines how many people are in each individual area and what the crowd density per square metre is. Directions of movement and changes in visitor flows can also be deduced from the time series. Heatmaps highlight areas with increased density.

A key distinction is the difference from traditional methods of personal identification: the AI is designed to recognise where people are – not who they are. According to the Fraunhofer IOSB, faces or other individual characteristics are not analysed. Only anonymous results are retained for the scientific system; clear images are not stored.

Research under real-world conditions

The role of Fraunhofer IOSB extends beyond the operational monitoring of this year’s Oktoberfest. The remit includes the scientific evaluation of crowd monitoring and the retrospective analysis of visitor flows. The findings are intended to help further develop the methods for future large-scale events.

The University of the German Armed Forces in Munich is also using the Oktoberfest as a real-world laboratory. The focus there is on models for simulating pedestrian flows. In the long term, the aim is not only to record current crowd densities but also to better forecast developments over the next few minutes and to simulate possible measures.

This heralds the next stage of development in crowd management: Instead of reacting solely to congestion that has already arisen, it may in future be possible to identify at an earlier stage where a critical situation is developing. For organisers and security officials, this would extend the time available to implement countermeasures.

However, this is still at the research stage. The decision on specific interventions remains with the responsible authorities and emergency services.

Crowd spotting remains a human task

Alongside technical analysis, Munich continues to rely on crowd spotting. Specially deployed personnel monitor the situation directly on site and are in constant contact with the festival management and security officers.

This second channel of information is crucial. Whilst video technology can capture crowd densities and movements over a wide area, on-site observers can assess behaviour, mood, local characteristics or the causes of crowd congestion. Both perspectives complement one another.

Communication has also been refined. In the event of critical visitor flows, pre-prepared text modules are available which can be used for loudspeaker announcements in several languages. The new broadcast booth in the security control centre supports this process.

Consequently, alongside detection, greater emphasis is now placed on the time between assessing the situation, making a decision and communicating it.

Shorter response times in medical care too

The time factor also plays a role in the emergency response plan. In addition to the first-aid station in the service centre, which is staffed round the clock, a second station will be set up at the Familienplatzl in 2026. The city explicitly justifies this on the grounds of shorter response times in areas further away from the main station.

In addition, up to three emergency doctor response vehicles, 19 ambulances and two patient transport vehicles will be available. The main station has up to 20 observation beds. A mobile CT scanner is once again intended to help diagnose injuries more accurately on site and avoid unnecessary transfers to hospitals. For the first time, two laboratory devices will also be used on the festival site to directly measure various blood parameters.

The principle is similar to that of crowd management: it is not just the available resources that matter, but also how quickly they can be deployed effectively at the right location.

Vehicle protection and drone defence

Alongside these measures, traditional security measures remain in place. Outside the inner security cordon, additional measures are being implemented to prevent vehicle-ramming attacks. Among other things, Schwanthalerstraße, west of Theresienhöhe, will be closed off with vehicle security barriers during peak times.

The knife-free zone has also been extended. It encompasses not only the actual festival site but also areas between the inner and middle security perimeters.

Another key focus is protection against unmanned aerial vehicles. A no-fly zone, which also covers drones, is once again in force around the Theresienwiese. According to its own statements, the Munich police have a plan in place for the detection and defence against unauthorised drones. Details are not being published for operational reasons.

The security plan thus ranges from traditional perimeter protection and access controls, through video analysis and crowd management, to airspace surveillance.

Commentary: Security arises between detection and intervention

Technical developments in the security sector often focus on how reliably a system can detect an incident. At major events, however, this approach falls short.

A camera can detect an increasing density of people. AI can analyse data more quickly, visualise directions of movement and filter out anomalies from a multitude of images. However, what happens next is crucial to the actual security impact.

There is a chain of processes between detection, assessment, decision-making, communication and intervention. Its total duration determines whether a technical information advantage actually translates into an operational security benefit.

This is precisely why the new security control centre at the Wiesn is at least as interesting as the AI being used. Data is not generated in isolation, but is intended to converge where different stakeholders can establish a shared situational picture and coordinate measures. Crowdspotter complements technical monitoring, whilst pre-established communication channels are designed to relay decisions to the site more quickly, and additional operational structures shorten response times.

The implications extend far beyond the Oktoberfest. When it comes to video analysis, perimeter protection, control centre technology or critical infrastructure, the detection performance of a system should not be the sole criterion for evaluation. Equally relevant is the end-to-end intervention time: how much time elapses from the first reliable indication to an effective response?

AI can accelerate this process by consolidating information and highlighting relevant developments at an earlier stage. However, it is no substitute for a clear decision-making structure or sufficient available operational personnel.

Modern security technology therefore realises its full value where sensor technology, AI, situational management and intervention are planned as an integrated system. Ultimately, the crucial question is not only: How quickly can we detect a threat? but also: How quickly can we respond to it effectively?