Xingyao Robotics | AISORT
AISORT technology combines optical capture, spectral identification, AI recognition, high-speed separation and cloud feedback into one engineering architecture for modern recovery plants.
Industrial line-scan imaging running at up to 200 fps for fast-moving production lines.
High-speed short-wave hyperspectral capture covering around 200 spectral bands.
Reported on daily mixed-material recognition tasks using self-developed visual models.
High-pressure valve execution designed for fast industrial separation timing.
AISORT is built as a closed technical chain: optical modeling, feature extraction, real-time inference and execution planning. This is the structure that allows one platform family to serve multiple recovery scenarios.
The sensing layer combines industrial area cameras, 12K line-scan imaging, SWIR hyperspectral imaging, ToF 3D structured light and binocular capture, supported by internally developed illumination systems.
Self-developed visual inference models and model libraries allow AISORT to classify material, shape, color and application-specific categories while supporting multimodal fusion with spectral data.
Model deployment can scale from high-performance NVIDIA GPU systems to edge NPU configurations depending on the complexity of the feedstock and the economics of the line.
Execution systems include 8 bar high-pressure valve arrays, delta robots and gantry robots selected according to the physical characteristics of each target stream.
Recovery plants are constrained by unstable feedstock, limited space, inconsistent labor and the need for repeatable output quality. AISORT is designed to respond to those constraints with equipment combinations that are technically coherent and commercially relevant.
Tower and mini configurations reduce the space penalty of upgrading smaller and medium-size facilities.
Shorter baseline modeling cycles improve responsiveness when new material classes need to be recognized.
The platform chooses between valve discharge and robotic extraction based on object size, mass and handling complexity.
Technology is designed to extend across material recovery centers, compact bottle lines and future distributed sorting nodes.
These technology blocks are expressed differently across the AISORT portfolio depending on throughput, material complexity and plant constraints.
Built for high-throughput optical recovery where fast visual classification and stable line speed are the primary requirements.
View productA compact multi-stage architecture that translates the imaging and execution stack into a smaller-footprint deployment model.
View productBest suited where richer spectral signatures improve downstream purification and quality stability.
View productApplies multimodal fusion when feedstock instability makes single-sensor classification insufficient.
View productTalk to our engineering team about material complexity, target purity, line fit and the right AISORT platform for your process.
Request Technical Review