Robot guidance
Chicken leg detection
A FANUC Delta M3ia robot with a SCHUNK needle gripper picks each leg guided by a 3D camera and a color camera, and a pneumatic system flips it.
Industrial computer vision: quality inspection and robot guidance
Cameras and software so your line checks every part and, when there is a robot, the robot knows where to pick it. All connected to the PLC and HMI in a single project.
What industrial computer vision is
Industrial computer vision uses cameras and software so a machine can see the product and make a decision: accept it, reject it, sort it or tell a robot where it is. As an official FANUC robot integrator since 2017, we combine it with the PLC, the HMI, the line sensors and, when the project calls for it, the robot, because a camera only truly pays off when its result moves something in the plant.
What we solve with computer vision
Quality inspection and defect detection
The camera checks every part and the system detects defects automatically. Automated visual inspection improves quality and reduces rework, without relying only on manual visual checks.
Inspection on your current line, without a robot
You do not always need a robot. We add inspection cameras and software to the line you already have, connected to its PLC and HMI, so the system flags or rejects defective parts. If you later want a robot to act on the result, we integrate a FANUC robot.
Robot guidance and vision pick and place
When the part does not always arrive in the same position, the camera tells the robot where it is and how to pick it. That is how the system we developed with a FANUC Delta M3ia robot guided by a 3D camera and a color camera works.
FANUC robotsDetection and sorting
The software recognizes what is in front of it and classifies it. With banknotes, vision detects the denomination, position and condition of each one so the cell can arrange, count and record them.
Field trials
We also test outside the lab. In our blueberry detection and handling project with computer vision, we captured and validated data on a farm to tune the detection, trajectories and behavior of the FANUC robot.
Cameras, deep learning and classic control
The camera and software are chosen based on what needs to be seen and what has to happen next.
A 3D camera provides the position and height of each part, which is key when a robot has to pick it. A color camera adds what the shape does not show, such as differences in shade. In the chicken leg detection system we use both.
We have worked with different camera brands, and the camera brand we have the most experience with is Basler.
To recognize objects we use computer vision and machine learning models, such as neural networks and deep learning, combined with classic automatic control for precision tasks. We work with the brands your plant already uses, without locking you into one supplier.
What we integrate with vision
- Cameras
- Color and 3D cameras, depending on the part and process.
- Software
- Computer vision, neural networks and deep learning.
- Robot
- FANUC robots, such as the Delta M3ia and SCARA models.
- End effector
- SCHUNK grippers or custom-built tools.
- Controls
- PLC, HMI, sensors and line conveyors.
How a computer vision system works on the line
The camera is only the beginning. The result has to reach the PLC and the robot for something to happen.
1. Capture
A sensor signals that the part has arrived and the camera takes the image with the lighting that part needs.
2. Analysis
The software locates the part, compares it with what is expected and returns a result: position, type, or whether it is good or bad.
3. Action
The PLC receives that result and acts: the robot picks the part where the camera saw it, the bad part is diverted from the line, or the process stops and raises an alert.
4. Logging
The HMI shows the operator what is happening and every result is logged, which gives the process traceability.
That is why we design the vision system together with the control panel and, if there is a robot, with the robotic cell, even when we add it to a line that is already running.
Computer vision projects
Developments by our team where vision guides the robot or decides what to do with each part.
Cash management
Computer vision for banknotes
Detects the denomination, position and condition of each banknote so the robotic cell can pick, sort and record it. See it in cash management
Handling
FANUC SCARA robot with computer vision
From modeling and simulation to a working prototype: a FANUC SCARA robot with computer vision picks and arranges banknotes with minimal manual intervention.
See also our case studies and how we design and integrate robotic cells.
How we run the project
We do not publish fixed prices: the cost depends on what needs to be detected, the line speed and whether the vision system only inspects or also guides a robot. We quote after the technical assessment.
We serve you from our office at Miraflores 383, Torre Centenario, Floor 22, Santiago, Chile. You can also call us at +56 2 2634 0001.
- Assessment
- We review the part, what needs to be detected and how the line works today, and propose the scope and work plan.
- Data capture and testing
- We capture and validate data under real conditions to tune the detection.
- Development and integration
- Vision software, robot programming and connection to the PLC, HMI and sensors.
- Commissioning
- On-site installation, testing and tuning with your team.
- After-sales
- Program adjustments, recipe changes and maintenance.
Computer vision FAQ
Can't find your question? Write to us and a specialist from the area will reply.
What is an industrial computer vision system?
It is a combination of camera, lighting, software and a connection to the plant controls that lets a machine see the product and decide what to do with it: accept it, reject it, sort it or tell a robot where to pick it.
What is computer vision used for in a plant?
To inspect quality and detect defects automatically, to guide robots picking parts that do not always arrive in the same position, and to detect or sort products. We have used it to guide a FANUC Delta robot in chicken leg detection and to detect the denomination, position and condition of banknotes.
What kind of camera is used in computer vision?
It depends on what needs to be seen. A color camera distinguishes shades and surface marks; a 3D camera provides the position and height of the part, which is key to guiding a robot. Sometimes both are combined, as in our chicken leg detection system. We have worked with several camera brands, and the camera brand we have the most experience with is Basler, the industrial cameras made by the German manufacturer Basler AG.
Do you use artificial intelligence or deep learning?
Yes. We use computer vision and machine learning models, such as neural networks and deep learning, combined with classic automatic control when the task requires precision.
Can it be added to an existing line?
Yes. We connect the vision system to the PLC, HMI, sensors, conveyors and robots already running on your line, and we work with the brands your plant already uses.
Can you implement computer vision without a robot?
Yes. We can add computer vision inspection to a line that is already running, connected to its PLC and HMI, so the system flags or rejects defective parts without a robot. If you later want a robot to act on the result, we integrate a FANUC robot.
How much does a computer vision system cost?
It depends on what needs to be detected, the line speed, the type of camera and whether the vision system only inspects or also guides a robot. We do not publish fixed prices: we quote after a technical assessment of the process.
What does your line need to see?
Tell us which part needs to be inspected or picked and how it arrives today. We will propose the camera, software and integration with your robot or PLC.