LIPS launches edge accelerator for robotic vision with Nvidia Jetson AGX Xavier module
LIPS Corporation, a provider of 3D AI solutions, has launched the LIPSedge F110 3DxAI edge accelerator for vision processing applications, such as facial recognition, medical imaging detection and autonomous vehicles.
The company demonstrated the hardware platform at Embedded World 2023. The edge accelerator speeds up the processing of 3D point cloud denoise, stitching, and fusion algorithms. One of these algorithms, 3D point cloud denoising, eliminates unwanted noise from a 3D point cloud data set.
A point cloud is a collection of 3D points that represent the surface of an object captured by sensors. These sensors can generate noise data due to external factors like environmental conditions and motion blur.
“Processing multiple 3D camera streams in real-time has always been a big challenge in the industry. LIPS is able to provide a low-latency and scale-up edge acceleration architecture that not only speeds up image processing but also AI inference,” says Luke Liu, the CEO of LIPS Corporation.
According to the company, the LIPSedge F110 3DxAI edge accelerator works with the LIPS Corporation’s proprietary LIPSedge 3D Camera, allowing for deploying enterprise-scale vision processing applications, such as facial recognition, medical imaging detection and autonomous vehicles.
LIPS has claimed that the LIPSedge F110 3DxAI edge accelerator is one of the few hardware systems that combine x86 and Nvidia Jetson architecture to integrate AI into conventional industrial automation systems. The hardware device supports edge computing applications and Nvidia CUDA and OpenCL, which are general-purpose programming languages for the graphics processing unit.
“By enabling NVIDIA Jetson running PCIe end-point mode on x86, we are also providing a next-generation building block to drive new innovations in edge-AI,” Liu adds.
Processing depth information from a 3D camera for vision processing generally requires a lot of power. As a result, using even a single camera stream can be complex and costly to manage, requiring a dedicated IPC.
However, LIPS executives say the LIPSedge F110 3DxAI edge accelerator overcomes this challenge by permitting a single system to pre-process multiple 3D camera streams while performing AI inference simultaneously. The company says this process improves the efficiency and accuracy of 3D vision applications in various fields, such as industrial automation, robotic vision and logistics.
Article Topics
autonomous vehicles | computer vision | edge AI | facial recognition | LIPS | Nvidia
Comments