FAQs About Our Solutions and How They Work
Explore how Blue Sky Robotics' intelligent automation solutions can transform your operations. Below you'll find answers to common questions about our computer vision software, system integration and capabilities, deployment and support, and industry-specific features.
COMPUTER VISION & OBJECT DETECTION
What is computer vision, and how does it enhance robotic arms?
Computer vision gives robotic systems the ability to visually recognize, locate, and interact with objects in real time. This eliminates the need for hard-coded programming and enables robots to adapt to changing environments—ideal for automation in logistics, manufacturing, and food processing.
How does computer vision expand automation capabilities?
With computer vision, robotic arms can work in dynamic, unstructured environments by identifying varied objects, shapes, and orientations. This flexibility allows for tasks like automated picking, sorting, quality control, and handling in industries where item variation is the norm.
How is a new item trained using computer vision?
We first capture hundreds of images and videos of the object in various positions and lighting conditions. This dataset is then annotated and used to train a deep learning model that enables accurate object detection and reliable robotic manipulation.
How long does training take, and can it be outsourced?
Typical training takes 10–14 hours, including setup and overnight model training. To accelerate deployment, we offer the option to outsource data annotation and model development to trusted third-party partners.
Are there limitations to the types of items that can be trained?
Training effectiveness depends on the item’s size, shape, and graspability. The robotic arm's payload (e.g., 5kg/10lbs for the uFactory xArm 6) and end effector compatibility are also factors. Soft, irregular, or reflective objects may require special handling.
SYSTEM COMPATIBILITY & FLEXIBILITY
Is your vision software compatible with different robotic hardware?
Yes. Our software is robot-agnostic, meaning it can be deployed across a wide range of robots. Whether you're starting from scratch or upgrading an existing system, our solution integrates seamlessly with your hardware.
Can different grippers or end effectors be used?
Absolutely! We support a variety of end effectors and can help you select the right gripper for your specific application. We also offer guidance for third-party options to ensure compatibility and performance.
What kind of cameras are used in your vision systems?
We currently use Intel RealSense depth cameras for robust 3D object detection. We're continuously testing new camera technologies to improve performance in diverse lighting and operational environments.
PERFORMANCE & CAPABILITIES
What is the typical pick speed using vision-based object recognition?
Our systems currently operate at an average of 1 pick every 4.5 seconds, depending on object complexity and motion path. This equates to approximately 276 picks per hour under standard conditions.
Can the vision model train multiple items at once?
Yes. You can train on multiple items simultaneously. However, the more objects involved, the longer the data collection and training time will be. We recommend strategic batch training to maximize efficiency.
INTEGRATION & WORKFLOW
How does your system integrate with existing software or warehouse management systems?
Integration is typically achieved via API, allowing communication with your warehouse or factory management software. We tailor each integration to ensure smooth data flow and task automation.
Is there a user interface available for system management?
Yes. As part of our services, we offer custom user interface design and development tailored to your specific solution needs. Additionally, we’ll soon be launching Spacebar, our proprietary no-code platform that allows users to build, manage, and edit robotic missions without writing code. Spacebar is designed to make mission setup intuitive and flexible across a wide range of applications.
DEPLOYMENT & SUPPORT
What is the typical implementation timeline?
Robot shipping generally takes 4–6 weeks (sometimes up to 8). Software development and training are conducted in parallel, with on-site setup depending on the number of robots and your workspace configuration.
Do you provide maintenance and ongoing support?
Yes. We offer optional maintenance and support plans tailored to your system’s complexity and uptime requirements. Services include remote diagnostics, software updates, and on-site support as needed.
INDUSTRY-SPECIFIC CAPABILITIES
Do you offer food-safe or explosion-proof robotic solutions?
Yes, we offer robotic systems that are food-safe and explosion-proof according to industry standards. We also design with these conditions in mind from the start, and in cases where the optimal robot for the application isn’t inherently rated for these environments, we will integrate protective coverings and enclosures that meet the required safety or hygiene standards. This approach allows us to balance performance and compliance within your operational constraints.
Can barcode scanning be used alongside computer vision?
Yes. While many clients eliminate the need for barcodes with our object recognition system, barcode scanning can still be integrated when required for traceability or regulatory compliance.