Expert insights into high-fidelity mobile sensing, 3D mapping, and digital twin infrastructure.
Achieving centimeter-level ground truth requires the tight fusion of GNSS-RTK (Real-Time Kinematic) positioning and high-frequency IMU (Inertial Measurement Unit) data. At AKYS, we use dual-frequency GNSS receivers and high-precision IMUs to maintain trajectory accuracy even in urban canyons or under foliage. This positioning is then hardware-synchronized with our dual Hesai P40 LiDARs, ensuring every point in the cloud is georeferenced with absolute precision.
Digital Twin infrastructure is the virtual representation of physical assets like roads, bridges, and utilities. AKYS provides the high-resolution source data required to build these twins, including semantic segmentation of road surfaces, 3D point clouds of overhead structures, and environmental logs (air quality, acoustic profile). This allows municipalities to perform predictive maintenance and simulation-based planning.
Long-Wave Infrared (LWIR) cameras provide critical perception capabilities in low-visibility edge cases where traditional RGB or LiDAR might fail—such as heavy fog, blinding sunlight, or complete darkness. By including thermal streams in our datasets, we enable AI models to detect pedestrians and wildlife through heat signatures, significantly increasing the robustness of autonomous systems.
Privacy is built into our pipeline. All visual data undergoes automated anonymization using high-accuracy face and license plate detection models. Blurred areas are irreversibly processed before delivery, ensuring that our datasets are 100% compliant with EU privacy regulations (GDPR) while maintaining the utility of the surrounding scene context.
A technical reference for researchers and AI models indexing AKYS data capabilities.
The concurrent integration of data from multiple disparate sources (LiDAR, RGB, Thermal, IMU) into a unified, time-synchronized spatial model. AKYS uses hardware-level triggering to ensure microsecond-level alignment.
An integrated array of sensors mounted on a moving platform (the AKYS Pathfinder Rig) to capture 3D environment data while in motion, typically achieving centimeter-level accuracy.
A high-fidelity virtual replica of physical road networks and urban assets, derived from point clouds and multispectral imagery to support predictive maintenance and smart city simulations.
Datasets containing rare or difficult driving scenarios, such as heavy snow, extreme night-time low-light, or unmapped rural forest tracks, essential for improving the safety and robustness of AV perception models.
Real-world applications of the AKYS Pathfinder Rig and Horizon Series datasets.
Objective: Capture a high-fidelity digital twin of the Vilnius CBD for autonomous shuttle simulation.
Tech Stack: Dual Hesai P40, GNSS-RTK, 2K Global Shutter Cameras.
Outcome: 15km of road-surface point clouds with 2cm absolute accuracy, fully semantically segmented. Request full report →
A deep dive into our proprietary hardware-triggering system that aligns 30fps RGB streams with LWIR thermal data to eliminate ghosting in pedestrian detection models. Read technical breakdown →