Challenge

"In Company R's production line, robotic picking systems encounter obstacles due to the diverse properties of objects handled, ranging from varying weights, textures, to different orientations. The engineering team conducts routine tests involving visual inspections to assess and readjust the robots for diverse tasks. However, the absence of automated, real-time adaptability within the robotic systems results in inefficiencies, elevated error rates, and potential production delays. These issues directly impact the company's throughput and operational costs, posing significant challenges to seamless production.

Solution

Company R implemented an AI Visual Inspection Kit empowered by a Siemens IPC 520A, running on the potent NVIDIA Jetson Xavier NX GPU, seamlessly connected to Teledyne GigE cameras. This cutting-edge solution enables on-edge data collection and computation while leveraging TensorBox-optimized computer vision models. These models empower our robots with the remarkable ability to 'see' and dynamically adapt to the diverse range of products within our production line. The AI algorithms facilitate the robots' ability to self-adjust to different object characteristics, enhancing the picking process's speed and accuracy.

Project Highlights

Within a concise span of 4 months, the deployment of the AI Visual Inspection Kit onto our production line resulted in remarkable advancements. The AI-enabled robots showcased a heightened proficiency in accurately picking a diverse range of objects, leading to a pronounced reduction in errors and downtime. The system's real-time processing capabilities enabled instantaneous adjustments to the picking process, eliminating the necessity for manual recalibrations. As a result, company R reaps substantial cost savings and solidifies a robust competitive edge in the market, setting a new standard for efficiency and accuracy in our production endeavors.