Outsight, Advantech, and Intel Unite to Launch a Smart Cities Solution

Frequently Asked Questions

• What does a LiDAR and camera fusion setup actually look like at a smart city intersection?

  In a typical deployment, LiDAR sensors and cameras feed into a single edge PC. The LiDAR workload runs on CPU cores, producing 3D tracking of vehicles and pedestrians, while camera frames are processed in parallel on the integrated GPU for video analytics. The two streams are then merged into one live spatial scene. In the smart city solution developed by Outsight with Intel and Advantech, this split-processing architecture runs on compact, affordable hardware rather than requiring a dedicated AI accelerator for each sensor type. The result is a fused, real-time spatial view of an intersection that can support safety programs like Vision Zero, similar to how Outsight's SHIFT platform is applied at instrumented urban intersections such as those in the City of Bellevue.

• Why run LiDAR processing on a CPU rather than a dedicated GPU or FPGA?

  LiDAR point-cloud processing is geometrically structured work: detecting, classifying, and tracking objects in 3D space across a sparse but ordered data stream. That workload maps efficiently onto modern CPU cores with wide SIMD instruction sets, unlike the dense pixel-grid operations that favor GPU parallelism. Outsight's SHIFT platform is built around this principle, running its sub-50ms end-to-end pipeline on CPU-based edge hardware without requiring a discrete graphics card. Keeping LiDAR processing on the CPU leaves the integrated GPU free for camera-based video analytics, so a single compact edge box handles both sensor modalities. In the Outsight, Advantech, and Intel smart cities deployment, this architecture lets infrastructure-mounted sensors deliver real-time 3D situational awareness at intersections and high-traffic urban areas within a practical, deployable form factor.

• How does a smart city LiDAR deployment handle GDPR without anonymizing data after collection?

  LiDAR sensors emit laser pulses and measure surface returns: the output is a 3D point cloud of shapes and positions, not pixels. Faces, license plates, and biometric identifiers are physically impossible to record at any stage of the pipeline. Because no personal data enters the system, there is nothing to anonymize retroactively. GDPR compliance is therefore a structural property of the sensor physics rather than a downstream data-handling policy. Outsight's infrastructure-based approach to smart city deployments, including Vision Zero intersection monitoring for the City of Bellevue, is built on this principle: the Motional Digital Twin captures how people and vehicles move through a space, never who they are.

• What is Intel SceneScape and how does it relate to LiDAR software?

  Intel SceneScape is a multimodal scene intelligence software layer that performs sensor fusion, combining inputs from cameras and other sensors into a unified real-time spatial representation. In the smart cities architecture developed with Outsight and Advantech, SceneScape handles video analytics on camera frames using the integrated GPU, while the Outsight SHIFT platform handles the 3D tracking workload on CPU cores. The two outputs are then merged to produce a single live scene rather than two disconnected data streams. This division of labor allows each processing layer to operate within its native compute domain, with LiDAR-derived 3D geometry and camera-derived visual context arriving as complementary inputs to a coherent operational picture.

• How many sensors does a LiDAR-based smart intersection need compared to a camera-only setup?

  LiDAR's 3D field of view covers a wider area per unit than a flat 2D camera, because depth perception reduces the number of overlapping vantage points needed to eliminate blind zones. A scalable LiDAR fusion architecture, where multiple sensors share one point cloud, further reduces per-intersection sensor counts. Outsight applies this principle in its infrastructure-based deployments, such as the Vision Zero smart-city intersections in the City of Bellevue, where LiDAR sensors mounted in the infrastructure track pedestrians and vehicles across the full intersection volume without requiring the dense camera arrays a 2D setup would demand. The practical effect is lower hardware, cabling, and maintenance costs over the life of the deployment, though exact counts depend on intersection geometry and the specific sensor models used.

• What smart city use cases go beyond traffic counting with LiDAR at intersections?

  Beyond vehicle throughput measurement, infrastructure-mounted LiDAR at intersections can support automatic incident detection (stopped vehicles, debris, collisions), safe-distance monitoring between road users, vulnerable road user safety alerts for cyclists and pedestrians, and parking occupancy monitoring in adjacent areas. The same sensor network that counts vehicles can trigger real-time alerts when a pedestrian enters a lane during a red phase or when a vehicle is stationary in a live traffic lane. Outsight applies exactly this range of use cases through its SHIFT platform, which powers Vision Zero-aligned deployments such as the City of Bellevue smart-city intersections, processing 3D motion data in real time without any camera feed or biometric capture.