FAQS
Find answers to some of our most frequently asked questions about hardware and software. Have another question and don't see the answer here? Reach out to us with any other questions via email and we'll get back to you with an answer as soon as possible.
1. WHAT IS COMPUTER VISION, AND HOW DOES IT ENHANCE ROBOTIC ARM CAPABILITIES?
Computer vision enables computers to identify and understand objects and people in images and videos. By incorporating computer vision, robotic arms can recognize and interact with objects without extensive manual programming, making them more adaptable and efficient.
2. HOW DOES COMPUTER VISION EXPAND THE RANGE OF APPLICATIONS FOR ROBOTICS?
Computer vision broadens the potential applications of robotic arms by allowing them to recognize and manipulate objects in varied and changing environments. This capability extends their use beyond predefined tasks, making them suitable for industries such as manufacturing, logistics, and food service. In almost any scenario where robotic arms are utilized, vision can enhance the arms' capabilities.
3. HOW LONG DOES IT TAKE TO TRAIN A NEW ITEM USING COMPUTER VISION?
Training a new item typically takes around 10-14 hours from start to finish. This duration includes approximately 2 hours for pre-training work and 8-12 hours for actual training, which is often done overnight. However, the timeframe may vary based on the complexity of the item and environmental factors.
4. CAN TRAINING BE OUTSOURCED TO THIRD PARTIES FOR FASTER PROCESSING?
Yes, training can be outsourced to third-party services for labeling and training, depending on project requirements and timelines. This option can expedite the training process and ensure efficient utilization of resources, but it does come with additional costs.
5. WHAT ARE THE LIMITATIONS REGARDING THE TYPES OF ITEMS THE MODEL CAN BE TRAINED ON?
Limitations include the payload capacity of the arm (5kg/10lbs for the uFactory xArm 6), the graspability of objects, and compatibility with end effectors. Soft or irregularly shaped objects may require specific handling solutions to ensure successful manipulation.
6. HOW MANY OBJECTS CAN THE VISION MODEL TRAIN ON SIMULTANEOUSLY?
The robotic arm can train on as many objects as needed. However, the pre-training time will increase proportionally with the number of objects, so it's essential to consider the scope and requirements of the task.
7. IS IT POSSIBLE TO EXCHANGE GRIPPERS? WHAT ASSISTANCE IS PROVIDED?
Yes, grippers can be exchanged depending on available options. We provide assistance in selecting suitable end effectors, and we offer guidance on third-party options if necessary to ensure compatibility with specific use cases.
8. WHAT IS THE PICK SPEED OF THE ROBOTIC ARM EQUIPPED WITH COMPUTER VISION?
The pick speed varies depending on detection and movement speeds. Currently, it's approximately 1 pick every 4.5 seconds in terms of detection, translating to around 276 picks per hour at 13 seconds per pick, including detection and arm movement.
9. DO YOU OFFER BARCODE SCANNING CAPABILITIES?
Barcode scanning can be considered for certain systems based on the use case. However, it may become obsolete with the use of the vision system, which offers robust object recognition capabilities. When specific identification requirements exist, barcode scanning may be recommended and can be taken into account for the project design.
10. HOW DO YOU TRAIN THE ROBOTIC ARM ON A NEW ITEM USING COMPUTER VISION?
The training process involves capturing several hundred images and videos of the object, annotating the data, and training a deep learning model. This includes manual labeling and data augmentation to improve model accuracy, followed by testing and validation to ensure reliable performance.
11. HOW IS INTEGRATION WITH EXISTING SYSTEMS ACHIEVED, AND IT THERE A USER INTERFACE AVAILABLE?
Integration is typically done through an API from the client's system, such as a warehouse management system. While a user interface is available with our Mission Manager product, it may not be customizable for specific systems. The Mission Manager system facilitates task management for multiple robots, ensuring seamless integration and coordination.
12. WHAT IS THE TYPICAL PROJECT IMPLEMENTATION TIME, AND DO YOU PROVIDE UPKEEP AND MAINTENANCE?
Robot shipping takes 4-6 (sometimes 8) weeks, and in-house development can be completed within that time. Hardware implementation by our experts varies depending on factors such as the number of robots and the workspace layout. We do provide upkeep and maintenance services at an additional cost, dependent on the specific needs and requirements of each use case.
13. WHAT TYPE OF CAMERA IS USED FOR COMPUTER VISION?
Intel RealSense cameras are currently utilized for computer vision applications. However, other camera options are undergoing testing to ensure compatibility and optimal performance in various environments.
14. WHAT ARE THE PRICING OPTIONS FOR YOUR SYSTEM?
The hardware, including one arm with an end effector, typically costs around $10,000. The vision software costs approximately 1-2 times the hardware cost.