IN Brief:
- Mantara combines event-based vision sensors, embedded AI processing, sensor fusion, and fleet-management software.
- Individual pixels respond to changes within a scene instead of producing conventional video frames.
- Operational counter-drone coverage will continue to combine optical, radar, RF, and acoustic detection.
Prophesee has launched Mantara, an integrated drone-detection platform combining event-based vision sensors, embedded artificial intelligence, sensor-fusion software, and remote system management.
Rather than recording complete image frames at fixed intervals, Mantara’s sensor responds independently when individual pixels detect a change in brightness. Static background information is largely discarded, allowing moving objects to be represented with lower data volumes and microsecond-scale temporal resolution.
The first configuration combines Prophesee’s imaging technology with IDS Imaging hardware and the company’s Hearth software layer, which manages sensor fusion, configuration, over-the-air updates, and deployed fleets. A memorandum of understanding with Exosens is intended to extend future systems beyond the visible spectrum.
Field trials have focused on passive detection around airports, ports, power stations, industrial sites, prisons, stadiums, and defence installations. Because the optical system does not transmit radio-frequency energy, it can operate without revealing its location or depending on an active communications link from the target drone.
Event-based sensing is well suited to rapid motion, particularly when a small object crosses a scene faster than the exposure and frame rate of a conventional camera can capture cleanly. High dynamic range can preserve changes around strong backlighting, shadows, or abrupt transitions between bright sky and darker structures.
Lower data rates also change the embedded-computing requirement. A frame camera continually sends information about every pixel, even when most of the scene is unchanged, whereas an event sensor concentrates processing on temporal changes. That can reduce memory bandwidth, communications traffic, and the energy consumed by continuous inference at the edge.
Those gains do not eliminate the optical constraints surrounding drone detection. Fog, rain, glare, foliage, buildings, and low-contrast backgrounds can obscure a target, while birds, airborne debris, and moving machinery create competing events that the classification system must distinguish without generating an unmanageable alarm rate.
Range remains closely tied to optics, sensor resolution, atmospheric conditions, and the apparent size of the aircraft. A small quadcopter approaching head-on presents less lateral motion than one crossing the field of view, so detection performance will vary with trajectory as well as distance.
Layered counter-uncrewed-aircraft systems combine complementary sensors to manage those weaknesses. Radar can provide range and track information, RF equipment can detect command or video links, acoustic arrays can identify characteristic signatures, and imaging systems can confirm classification. Recent counter-drone and electronic-warfare systems from Aselsan use the same multi-sensor principle rather than depending on a single detection channel.
Mantara’s software layer will determine how effectively its optical data enters that wider architecture. Track correlation, timing, coordinate alignment, confidence scoring, and interfaces to command systems have to remain consistent when sensors differ in update rate, field of view, and measurement uncertainty.
Remote management introduces another set of engineering controls. Detection models and threat libraries need regular updates as aircraft designs, operating tactics, and local backgrounds change, yet the update path must be authenticated and protected against rollback, configuration corruption, or unauthorised access.
Secure boot, signed software, controlled administrator privileges, and a recoverable previous image are especially important when the same platform is deployed across several critical sites. A compromised fleet-management service could otherwise create a common failure point across installations that were intended to operate independently.
Prophesee has raised €20 million, led by Critical Path Ventures, to support Mantara and the wider commercial development of event-based vision. The funding arrives as infrastructure operators and governments expand counter-drone procurement following repeated incursions near airports, military facilities, energy assets, and public venues.
The platform also reflects the convergence between defence sensing and industrial machine vision. Event-driven cameras have already been applied to robotics, inspection, autonomous systems, and high-speed motion analysis; drone detection adds long-range observation, adverse weather, small targets, and safety-critical classification to the same underlying electronics.
Operational performance will be judged through detection probability, identification accuracy, false-alarm rate, latency, and availability under difficult backgrounds and weather conditions. Mantara provides a low-data, passive optical layer, while dependable site protection will continue to rely on how effectively that layer is fused with the other sensors and command systems around it.
