Combat cognition AI
Senshi
Senshi supports the people interpreting the operating picture and video. It raises alerts and explains what the evidence supports, while an authorized human makes the decision.
Picture and video capabilities on the station
Available in the demonstration
An extra view of the picture.
The current software combines on-station video detection with assessment of the operating picture. Its interface separates observed information from inferred assessment and makes the limits of the record visible.
Video detection
Detect and count vehicles, people and bicycles in the demonstrated video workflow.
Counts and flows
Review aggregate activity and crossings without identifying people.
Lost-contact alerts
Bring a watched track's unexplained disappearance and last known information back to the operator's attention.
Plain-language explanation
Read explanations grounded in the engine's recorded picture, with unsupported answers identified.
Decision support
The engine assesses. A human decides.
Senshi's role is to alert and explain. Authorization remains a human action under HANKO. In the demonstrated video workflow, processing happens on the station, video is not recorded, and people are counted rather than tracked by identity.
Research and development
A defined roadmap for richer context.
Future development focuses on behaviour recognition, audio-event detection and combined text, audio, image and video understanding. Each area will build on the current picture and video workflow with its own evaluation.
A closer look
Questions and answers.
Does Senshi authorize actions?
No. Senshi provides assessment, alerts and explanations. Human authorization is handled through HANKO.
What has been demonstrated?
The running station demonstrates video detection, counts and flows, operating-picture assessment, lost-contact alerts and plain-language explanation. The video demonstration uses stock footage replayed as a live feed.
How does the video workflow handle people's data?
The demonstrated workflow counts people and aggregate activity without tracking people by identity. Processing happens on the station, and video is not recorded.
Connected software
See the software
Bring your questions to the station.
Discuss a demonstration, the current evidence and your integration requirements.