EVENTS – ReliablE in-Vehicle pErception and decisioN-making in complex environmenTal conditions
As Europe moves toward the large-scale deployment of Connected, Cooperative and Automated Mobility (CCAM), ensuring that automated vehicles can reliably perceive their surroundings and make safe decisions in every situation remains a central challenge. The EVENTS project is addressing this challenge by developing a new generation of perception, prediction, decision-making and fail-safe control technologies designed to operate in the most demanding real-world environments.
Although EVENTS is still undergoing its final testing phase, the project has already delivered significant technological advances that strengthen the foundations for safe and trustworthy automated driving in Europe.
Building a Comprehensive Foundation for Automated Driving
EVENTS brings together advanced sensing technologies, artificial intelligence, cooperative communication, behavioural modelling and vehicle control into a unified architecture for automated driving. The system integrates data from cameras, LiDAR, Radar, GNSS/INS, high-definition maps and V2X communication to form a rich, multi-layered understanding of the driving environment.
A defining aspect of the project is its emphasis on self-assessment, enabling the perception system not only to detect objects but also to evaluate the confidence of its own outputs and recognise when conditions may compromise performance. This function is being applied across scenarios that include interactions with vulnerable road users, occluded intersections, temporary road layouts, adverse weather and sensor disturbances. It marks a key step toward transparent, reliable and safe automated driving.
Alongside perception, EVENTS is developing advanced decision-making tools, trajectory planning strategies and robust control mechanisms that allow the vehicle to navigate complex environments and respond safely when uncertainties arise.
Strengthening Environmental Awareness in Challenging Conditions
The project has achieved substantial progress in developing multi-sensor perception pipelines capable of supporting automated driving in highly dynamic and diverse environments. These pipelines provide the foundation for urban interaction with pedestrians and cyclists, cooperative driving in roundabouts, moving through roadworks, merging onto highways and navigating areas with poor visibility.
Machine learning also plays a central role in detecting small or unusual objects, such as road debris, and determining whether they can be safely crossed. To support this, dedicated datasets have been collected to improve performance in adverse weather and low-visibility scenarios.
Predictive intelligence is another cornerstone of EVENTS. Algorithms capable of forecasting the future paths of surrounding vehicles and vulnerable road users enhance the vehicle’s ability to understand and anticipate complex interactions. When supported by V2X communication, this predictive capability becomes even stronger, allowing infrastructure and nearby connected vehicles to contribute to a shared understanding of the environment.
Transforming Environmental Understanding into Safe Driving Actions
In parallel to perception, EVENTS is advancing the decision-making and motion-planning components that translate environmental understanding into safe and effective driving actions.
A versatile motion-planning framework now enables the vehicle to generate smooth, efficient and dynamically feasible trajectories across a wide range of situations. These trajectories are produced using a mix of established curve-based techniques and optimisation-driven methods, ensuring adaptability to both structured and unstructured road layouts.
Decision-making strategies within the project address several demanding traffic scenarios: unprotected intersections with or without infrastructure support, interactions with vulnerable road users, re-establishing a platoon formation after navigating a roundabout, choosing lanes in roadworks with temporary boundaries and predicting driver behaviour for enhanced cruise-control functions. In several cases, the project makes use of reinforcement learning and deep-learning-based intention prediction to support more natural and anticipatory automated driving.
Together, these tools allow the vehicle to manage both everyday manoeuvres and unexpectedly complex or ambiguous situations.
Ensuring Safety When Conditions Become Critical
An essential part of EVENTS is the development of robust safety measures that govern the vehicle’s behaviour in emergency or degraded conditions. These measures ensure that when perception becomes uncertain, decisions cannot be executed reliably or the human driver does not respond to a take-over request, the automated system can always guide the vehicle to a safe state.
The project has therefore designed a dedicated framework for evaluating risk, selecting appropriate minimum-risk manoeuvres and applying controlled braking or evasive actions. This framework ensures that safety remains the overriding priority, even when the operating conditions become highly challenging.
From Research to Reality: Integration in Prototype Vehicles
EVENTS has reached a major milestone with the integration of its complete software stack into several prototype vehicles equipped with multi-sensor suites and drive-by-wire systems. This integration encompasses perception, prediction, planning, control and cooperative communication components.
All project scenarios – from dense urban interactions to cooperative manoeuvring, temporary road layouts, highway merging and emergency actions – are now running on real vehicles. Early integration tests show that data flows smoothly between all modules, algorithms operate in real time and the vehicles can execute planned trajectories with high accuracy. These achievements confirm the system’s readiness for the extensive validation phase ahead.
A Safety-Driven Path Toward Deployment
To support the ultimate goal of safe automated driving, EVENTS is developing a rigorous methodology that follows the principles of the Safety of the Intended Functionality (SOTIF). This approach identifies potential triggering conditions, defines safety goals and derives measurable requirements for the vehicle, subsystems and individual modules.
Scenario-based testing, simulation, statistical analysis and structured safety reasoning are being combined to evaluate how the system behaves across its operational design domain. This methodology will guide the final verification activities and ensure that the technologies developed in EVENTS can be trusted in diverse and unpredictable real-world environments.
Looking Ahead
With its core components now integrated and functional on demonstration vehicles, the EVENTS project is entering its final phase of full-scale testing and validation. The progress achieved so far highlights the potential for automated driving systems that can understand complex environments, monitor their own performance and act safely even when faced with uncertainty.
Through its strong focus on robustness, cooperation, predictive intelligence and safety-driven design, EVENTS is contributing to Europe’s ambition to deploy reliable, trustworthy and human-centred automated mobility solutions.
