Lagging detection
Manual scoring is subjective and slow; simple sensors miss fine or rare events and struggle with multi-animal interaction.
Closed-loop Control
Computer vision, edge compute, and flexible trigger logic — track pose in real time, recognize custom behaviors, and drive optogenetics, electrophysiology, and recording devices for detect-to-intervene closed loops.

WHY CLOSED-LOOP
Passive observation is no longer enough — lagging detection, fragmented systems, rigid paradigms, and brittle algorithms limit closed-loop timing and precision.
Manual scoring is subjective and slow; simple sensors miss fine or rare events and struggle with multi-animal interaction.
Tracking and stimulation are often separate — sync and latency are hard to control, limiting closed-loop timing.
Many systems lock to one paradigm — hard to move from position tracking to pose or custom behaviors like maternal care.
Classical vision is fragile to lighting, occlusion, and individual variation — weak on pose-defined complex behaviors.
CORE CAPABILITIES
Real-time monitoring, intelligent analysis, and precise closed-loop feedback in one chain — from pose to custom behavior to external device triggers.
High-precision position and pose for one or many animals — full timelines and spatial conditions like social distance.
Compatible with DeepLabCut, YOLO, and custom models — recognize user-defined complex behaviors with stronger robustness.
Low-latency rule engine outputs TTL sync signals on condition combinations — drive optogenetics, stimulators, and recorders.
PRODUCT FEATURES
AI precision, a closed-loop core, and open interfaces — from standard rodent paradigms to custom mid/large-animal setups.
Deep learning for keypoints, pose, and behavior — stronger than classical vision; load your own trained models.
Minimize detect-to-trigger delay; rules span simple to complex condition trees for diverse designs.
Standard mouse/rat support; customizable for hamster, guinea pig, monkey, dog, and arenas, mazes, home cages, social boxes.
Standard event-sync triggers connect to mainstream optogenetics, electrophysiology, imaging, and other I/O.
Core control software runs on Windows and Linux — fits existing lab platforms and workstations.
Replace manual watch and triggers — higher throughput, objectivity, and reproducibility for complex assays.
USE CASES
One detect → rule → TTL chain — either drive optogenetic intervention or stamp neural recordings with precise behavior timestamps.
Intervention loop
Track multiple mice in an open field and their pairwise distance. When distance falls below a set threshold (e.g. 5 cm), the rule engine fires TTL via the event sync trigger to drive optogenetic stimulation of a target region — probing causal effects on social approach or avoidance, with live distance, track masks, and trigger state on screen.

Sync recording
An AI model monitors parturition in real time. On key events such as first or second pup birth, the system immediately emits TTL to an external calcium / imaging recorder — writing precise behavior timestamps onto the neural timeline for post-hoc alignment of delivery-related activity and behavior sequences.

DEVICE LINKAGE
When behavior conditions are met, drive external devices instantly — stimulation and recording share one trigger path.
Behavior-contingent optical stimulation for causal circuit intervention.
Timed or conditional electrical stimulation aligned to behavior windows.
Auto-start or TTL-mark sync recordings — align neural signals to behavior timestamps.
Standard interfaces for additional stimuli, reward/punishment, or environment control.
WORKFLOW
From scene capture to live closed loop and event logging — one software stack for setup, rules, monitoring, and recording.
Capture scene video to train or validate tracking and behavior models.
Deploy workstation and cameras, configure the arena, connect and calibrate external triggers.
Define closed-loop rules (logic, signal type/duration) and video/event logging options.
AI parses pose or target behaviors live; on match, send control signals while the UI shows detection and trigger state.
Auto-save experiment video, trigger timestamps, and key parameters for replay and publication.
One GUI for setup, model loading, rule editing, live monitoring, and light visualization — no tool hopping.
SPECIFICATIONS
Edge compute + open triggers + customizable species/scenes — modular fit to existing lab gear.
Explore AI behavior recognition, rule configuration, and TTL device linkage — upgrade passive observation to intervenable, reproducible closed-loop experiments.