GPU-accelerated Optical Sensor Simulation - Simulating a Network of Optical Sensors Utilizing GPU-acceleration

Sammanfattning

Laser triangulation sensors are widely used in industrial measurement systems, where multiplesensors continuously acquire geometric data and transmit it to a host for calibrationand analysis. Prototyping such systems is costly and time-consuming, as physical sensorsrequire specialized hardware, precise alignment, and dedicated network infrastructure.This thesis presents a proof-of-concept framework for virtual laser triangulation sensorsthat can be used as a substitute for physical prototypes during system development andtesting.The work consists of two main components: a mathematical simulation for generatingsensor-like measurements, and a network layer that enables the virtual sensors to communicatewith the host software exactly as real devices do. The simulation computesraypolygon intersections to emulate the measurement process of a triangulation sensor.A naïve CPU version and a GPU-accelerated version were implemented, followed by acustom CUDA kernel based on Cramers rule for solving large batches of independent2 × 2 systems. Profiling and roofline analysis show that the custom kernel achieves severalorders of magnitude higher performance compared to both the CPU implementationand high-level GPU libraries such as cuSOLVER.The network interface is implemented using UDP communication and a virtual Wire-Guard network, allowing each virtual sensor to appear indistinguishable from a physicalone to the existing configuration software. This enables seamless hardware-in-the-loopstyle testing without modifications to the host system.The results demonstrate that virtual laser triangulation sensors can generate realisticmeasurements at rates significantly higher than required for real-time operation, creatingroom for future improvements in physical accuracy and noise modeling. The frameworkestablishes a foundation for scalable virtual prototyping of optical measurement systemsand shows that highly specialized GPU kernels can dramatically accelerate small-matrixcomputations commonly found in geometric simulation workloads.