Technology summary / value proposition
The technology is a roadside perception and communication system that helps automated vehicles react in time to pedestrians, cyclists, scooter riders, people with dogs and other vulnerable road users who are not equipped with special digital devices and may be outside the vehicle sensor field of view.
A roadside unit with a thermal camera detects a person or cyclist in a blind spot, determines their approximate location and sends this information to the automated vehicle in real time. The vehicle can reduce speed, stop or adapt its driving trajectory before the person becomes visible to its own cameras or laser sensors.
The main benefit is earlier warning in dangerous situations where the vehicle’s own perception is not sufficient. This reduces collision risk and extends the range of situations in which automated vehicles can operate safely.
Full description: business and innovation perspective
Challenge
Automated vehicles rely heavily on their own cameras, laser scanners and radars. These sensors work well when an object is visible, but they cannot see through parked cars, buildings, fences, road works equipment or vegetation. This creates a safety gap: a pedestrian, cyclist or scooter rider may enter the road from a hidden position, leaving the vehicle too little time to react.
This is an important problem for the deployment of connected and automated mobility in cities, suburbs and on rural roads. The risk is especially high near pedestrian crossings, parked cars, junctions with poor visibility, road works zones, parking areas and exits from factories, farms or residential areas.
Existing alternatives
Currently, similar safety problems are usually addressed with:
- better sensors on the vehicle itself;
- warning signs or traffic calming measures;
- a safety driver in the vehicle;
- standard traffic cameras that are not directly integrated with the vehicle decision-making system.
These approaches have limitations. Vehicle sensors still cannot see people behind obstacles. Road signs do not provide real-time information. Human supervision limits scalability. Standard traffic cameras are usually not directly connected to automated driving systems.
EDI solution and uniqueness
The EDI solution adds intelligent perception to roadside infrastructure. Instead of relying only on the vehicle’s ability to detect everything by itself, the roadside unit monitors safety-critical blind spots and sends only relevant warnings to the vehicle.
Distinctive advantages of the solution:
- the thermal camera can detect people in low-light and rainy conditions;
- data processing takes place locally on the roadside unit, so video frames do not need to leave the device, preserving people’s privacy;
- the system is designed for people who do not carry any special digital device communicating with cars or infrastructure;
- the vehicle receives the warning before its own sensors can detect the person;
- communication between the roadside unit and the car can take place through a 5G mobile network or through a direct communication channel;
- the solution has been tested in realistic demonstrations, including on an open road in Ādaži.
Technology readiness level
Current technology readiness level: TRL 7.
The full solution has been demonstrated in a relevant operational environment, including the closed Biķernieki track and an open rural road with low traffic intensity. Wider deployment still requires additional testing in dense traffic, complex weather conditions and long-term roadside operation.
Intellectual property status
EDI know-how in roadside perception, local artificial intelligence processing, movement trajectory analysis and integration with automated vehicle behaviour modules.
Projects
The technology was developed and demonstrated in the Horizon Europe project AUGMENTED CCAM: Augmenting and Evaluating the Physical and Digital Infrastructure for Connected, Cooperative and Automated Mobility deployment.
Project website:
EDI project page:
Technical specification
Operating principle
The system consists of a roadside unit and software modules in the automated vehicle. The roadside unit includes a thermal camera, a video processing platform, a positioning module and communication modules supporting both the 5G mobile network and direct communication. It observes a road area that may be hidden from the vehicle, detects vulnerable road users, determines their location and movement, and sends a warning message to the vehicle.
The vehicle receives the message through the mobile network or direct short-range communication. The driving software analyses the received information and, if the situation is assessed as dangerous, can reduce speed, stop or change lane.
Test results
The technology has been tested in component tests, integrated tests and demonstrations. The tests included pedestrians, cyclists and people with dogs. The system was demonstrated in sunny, cloudy and rainy weather. In the Ādaži demonstrations, the roadside unit detected road users outside the vehicle’s direct sensor field of view, and the vehicle reacted by reducing speed or stopping.
Key technical parameters
| Parameter | Value / description |
| Roadside perception | Thermal camera for detecting pedestrians, cyclists and larger animals |
| Local processing | Artificial intelligence model on the NVIDIA Jetson platform |
| Positioning | Roadside unit self-localisation with GNSS; detected road users are assigned world coordinates |
| Communication | Mobile network and direct short-range communication from infrastructure to vehicle |
| Vehicle integration | Autoware-based behaviour module for speed reduction, stopping or lane-change decisions |
| Tested road-user types | Pedestrians, cyclists, people with dogs; children, scooter riders and motorcyclists require additional testing |
| Detection performance | 88% correct classification on test video |
| Position accuracy | Average geo-coordinate error of 0.73 m |
| Communication reliability in tests | 100% message reception in both tested communication channels |
| Integrated response | The vehicle reduced speed in 100% of integrated tests; in 90% of relevant cases braking started within 1 second of detection |
| Current prototype limitations | Long-term robustness in harsh outdoor conditions and energy efficiency need improvement |
Cooperation model
EDI sees several possible routes for technology transfer:
Licensing
A partner may be granted rights to use selected EDI know-how, software components or system design elements in roadside safety products.
Contract research and cooperation
Joint development with road operators, municipalities, mobility technology companies or automated vehicle developers to adapt the system to a specific road environment, communication channel, vehicle platform or safety requirement.
Pilot deployment
Installation and evaluation of the system at high-risk road sections, such as pedestrian crossings, junctions with poor visibility, road works zones, parking areas or exits from factories, farms and residential areas.
Assignment or broader commercialisation
Transfer of selected intellectual property or broader commercialisation may be assessed after clarifying the intellectual property status and partner needs.
