AI & Machine Learning

Autonomous Adaptable Robot System (AARS), pioneering 3D vision Al-driven robot path generation, facilitates rapid, no-code robot programming and complex surface trajectory planning capability. lt breaks free from traditional teaching and offline programming constraints, enhancing the adaptability and intelligence of robots, improving user-friendliness, and achieving flexibility, convenience, and high efficiency. lt empowers individuals to become proficient in robotic automation without needing advanced training or prior experience.

Cambrianビジョンシステムは、高コスト効率、高速、高安定性を誇り、ばら積みピッキング、ピック＆プレース、マシンフィード、フィーチャーロケーション、アセンブリやスーパーマーケット、物流倉庫における利用を含む自動化及び製造関連のタスク向けに開発されました。 本システムはAIをベースにしています。正確な計測を行うために、AIモデルはロボットアームまたは作業スペース上部に設置が可能なステレオカメラが捉える画像データを使用します。対象アイテムの計測にストラクチャードライトや赤外線などの補助照明は必要ありません。そして様々な外光条件下で高い安定性を発揮します。このAIモデルは、200ミリ秒以内に、正確な部品位置と適切な把持位置を計算することができます。 　本システムは、一般的なプラスチック部品や金属部品のほか、透明、薄型、あるいは非常に光沢のある金属部品など、さまざまな種類の部品をピッキングすることができます。標準装備のカメラモジュールでは、1ミリ以下から20センチの立方体に収まるサイズまでの部品をピッキングすることができます。大型部品用には、少し大きなカメラモジュールをオプションでご用意しています。 本システムは素早い取り付けが可能で、UR準拠のグリッパーへの様々なアタッチメントプレートをご用意しています。

Inbolt is the leader in Dynamic Vision-Guidance Systems for Robots. Trusted by brands like Stellantis, Toyota, Beko and Ford, our system combines real-time 3D vision and AI to give robots the intelligence to adapt to their environment on the fly, whether it be moving lines, or cluttered pallets. Inbolt delivers guidance that’s faster, more flexible, and easier to scale than anything else on the market. Our robot-mounted vision system offers the fastest refresh rate in the industry, enabling real-time part localization and trajectory adjustment, even in poor lighting and dynamic environments. We support switching between over 80 part references within a single program, and our proprietary AI is trained directly on CAD models on-premise in under 10 minutes. There’s no need for jigs or constrained setups, and our system integrates natively with Universal Robots, requiring no workflow disruption.

Designed from the ground up to enable software defined autonomous machines, you can develop and deploy AI enabled robotics applications at scale with the powerful NVIDIA® Jetson AGX Xavier™ Developer Kit. It’s capable of running modern, advanced neural networks and other AI workloads to develop and deploy end-to-end AI applications. Plus, it delivers up to 32 TOPS of compute performance and can operate in as little as 10 W. The Jetson AGX Xavier Developer Kit helps you speed innovation. As part of NVIDIA’s world-leading AI platform, it’s integrated with NVIDIA tools and workflows that let you train neural networks faster, which means more iterations and better accuracy. It’s also supported by NVIDIA CUDA®, cuDNN, and TensorRT™ software libraries, as well as our JetPack™, DeepStream and Isaac SDKs. This helps you boost performance while reducing development cost and effort.

Locator Studio is ideal for vision-guided robotic applications based on referenced position: picking objects from a tray, conveyor belt, rack, or to locate components, fixtures for precise placing, screwing, and assembly. It serves well for picking parts from a non-collision environment, handling structured and semi-oriented parts, or de-racking applications. Additionally, it is ideal for simple depalletization tasks with its pre-trained neural networks or picking whole layers using the layer-based localization feature.

MIRAI（マイクロプシ・インダストリーズ社製）は、産業用ロボットが位置、形状、動きのばらつきに対応できるようにするビジョンベースの制御システムす。人工知能を使用したMIRAIコントローラーは、個々の測定値を出力するのではなく、ロボットの動きを直接リアルタイムで生成します。MIRAIは、組立、試験・測定、機械加工、把持、輪郭追従（分散など）などに活用できます。MIRAIは、タスク内の複数のステップ（例えば、可変位置からの把持や、形状が変化する組立など）に使用できます。また、新しいタスクのために柔軟に再教育することも可能です。スキルはプログラムされるのではなく、人間の実演を通して数日間で訓練されます。

SIEMENS offers off-the-shelf Advanced Robotics applications powered by AI, bridging the gap between technology and real-world problems, allowing you to scale new business opportunities. SIMATIC Robot Pick AI is a pre-trained vision software that leverages industrial-grade generative AI to make precise and reliable decisions for suction gripper pick poses (X, Y, Z, and approach angle) on any item. It excels in autonomous bin-picking, pick-by-robot applications, and beyond, without the need for user-specific item training with CAD models. Compatible with any Universal Robots Cobot, SIMATIC Robot Pick AI ensures superb and easy integration into the Siemens SIMATIC PLC landscape. Experience an effortless and unique integration of your Cobot robot logic by utilizing the SIMATIC Robot Library, which is based on the Standard Robot Command Interface. SIMATIC Robot Pick AI supports a range of 3D cameras (RGB-D) from various suppliers, from low-cost to high-definition. SIMATIC Robot Pick AI is especially suitable for items in: General Merchandise / eCommerce Grocery Pharmaceutical AI-powered robots are transforming warehouse operations by supplementing manual, repetitive, and cognitively demanding tasks, unlocking human potential for more creative and valuable work. Picking and placing tasks have always been among the most cost- and labor-intensive activities in warehouses and production operations. The bin-to-bin picking process, crucial for order fulfillment, leads the way in average labor demand growth. Contact our team today to revolutionize your solution portfolio and operations with AI-powered robotics. Take the first step toward unmatched efficiency and innovation.

Groundlight is a complete machine learning-powered computer vision solution for industrial inspection, anomaly detection, and much more.  The Groundlight Compute Module plugs directly into any e-series or CB-series robot and works with most industrial cameras including USB, GigE, RTSP, and many other types of streams including some depth cameras.  The URCap allows you to quickly add image queries to your programs for anomaly detection, setup verification, part identification, and more. Optimized for high-mix environments, setup verification, and part identification. Groundlight is low-cost, easy to operate, quick to deploy and thrives in complex visual environments where conventional computer vision capabilities fall short.  Using Groundlight's unique cloud-based AI model orchestration and escalation technology, the solution starts working from the first query and learns from both local and remote operator input as necessary. The Groundlight URCap can also generate inspection reports in addition to extensive statistics and performance analysis available through the Groundlight app. Get started now with cutting-edge and future-proofed computer vision capabilities today.

The industrialized 3D camera robobrain.eye, as part of the robobrain® technology was designed to help eliminate one of the most complex challenges in the automation industry: human-like hand-eye coordination with robots. Having the right tool which can see, recognize, and grip objects can all be achieved using this smart technology. The combination of robobrain®, robobrain.eye and suitable AI skills allows the coordination and separation of even overlapped objects for more flexibility and efficiency in the process. The industrialized 3D camera can either be mounted onto the robot arm itself or be placed above on the stations structure. robobrain® is the brain, that gives industrial robots human intelligence and unites Artificial Intelligence, vision and smart skills. With robobrain®, robots gain a level of intelligence that transcends previous limitations. Gone are the days of tedious and time-consuming learning processes; robobrain® empowers robots with the ability to learn and adapt swiftly. Its universal compatibility sets it apart from traditional control platforms – robobrain® operates independently, making it compatible with a wide range of different robotic components. With the URCap robobrain® can be easily combined with any UR Cobots and automation solutions can be implemented time and cost-effective.