The Libellula v10.3 IP module represents a breakthrough in vision processing technology. Designed for Event-Based Vision (AER) sensors, it processes asynchronous light intensity changes in real-time—unlike conventional frame-based cameras that capture snapshots at 30-60 Hz.
Result: Reaction times 1,000× to 10,000× faster than traditional vision systems, using a fraction of the power—no GPU required, only efficient fixed-point logic.
The Libellula v10.3 module addresses critical needs in the rapidly expanding Edge AI ecosystem, enabling breakthrough performance in:
Challenge: Latency causes motion sickness in VR environments.
Solution: Sub-microsecond eye-tracking and hand-tracking enable immediate rendering adjustments, eliminating lag-induced discomfort.
Power Advantage: Activity-scaled power consumption extends battery life in portable devices.
Challenge: Standard cameras blur fast-moving objects on production lines.
Solution: Tracks motion without blur at high speeds, enabling defect detection and counting for pills, components, or products moving at extreme velocities.
Challenge: Vehicles travel several meters during standard camera frame delays.
Solution: Instant detection of movement enables faster reaction to pedestrians and obstacles, potentially reducing stopping distance by meters.
Challenge: GPS-free navigation and stable hovering require precise motion tracking.
Solution: Optical flow data (dir_x, dir_y outputs) enables Visual Odometry and SLAM for indoor navigation and stable flight control.
Optimized for "SWaP" (Size, Weight, and Power) critical applications where reaction time can mean the difference between mission success and failure:
At Mach 5+, standard 60Hz cameras miss targets entirely between frames. The 30ns latency enables terminal guidance for high-velocity munitions and interceptors.
When GPS is jammed, autonomous systems use optical flow to navigate by tracking ground texture. This lightweight IP enables small UAVs and loitering munitions to operate autonomously.
Vehicle APS systems must intercept incoming RPGs in milliseconds. Ultra-low latency is critical for calculating intercept solutions within the narrow engagement window.
Event cameras excel at detecting rapid temporal changes like muzzle flashes. This IP instantly filters and pinpoints shooter locations (x_hat, y_hat coordinates).