In 2015, a worker was accidentally killed by an industrial robot in a Volkswagen plant in Germany. According to the Telegraph, the 22-year old technician was instantly killed as the robot’s arm struck his body and pushed him against a metal plate.
According to a statement from Volkswagen, the robot did not breakdown. The robot was operating inside of the safety cage when the technician mistakenly stepped inside of this cage. Accidents like this are rare, but when they happen they draw a lot of attention.
After this accident and another one similar in a car factory in India, some autonomous technology vendors re-directed most of its efforts to a new generation of safe collaborative robots, the Cobots.
The advancements in sensors, video cameras, AI, and computer vision technology are making these collaborative robots highly aware of their surroundings. Having extra visibility allows these next-gen robots to be more effective and safe for some type of factories.
Cobot: A Safe Robot for the Masses
“Collaborative robots in some size, will democratize access to robots. It used to be that we would only see robots in very large enterprises, the automotive or the aerospace. Now through collaborative robots, it is made possible for various mom and pop’s shops to adopt a robot. It is simple enough to program… they can install it, and they can use it within half an hour”, said Henrik Christensen, Director of Robotics of the University of California San Diego.
To make these collaborative robots accessible to everybody, the autonomous industry is working hard to keep up with safety standards while still producing efficient robots.
Collaborative Robots or Cobots are designed to share workspace with humans. They are completely autonomous systems that can be used for a wide arrange of applications in manufacturing, research, labs, etc. Cobots can pick and place items, pack them, inject molten into a mold, carve wood (CNC), polish, take analysis, and a lot more.
These cobots are usually labeled as safer than big industrial robots. But to be able to work alongside humans and without safety cages, there are some requirements that let cobots to be used as safe workplace collaborators. Some of these are:
- Monitored stop.
- Keeping track of motion and speed.
- Limit power and weight-lift capacity.
- Guide them by hand.
Cobots are power-enforced limited autonomous systems. They are designed to be small and not very powerful. The payloads that these limited robots can take are generally around 10 kgs or under. And if they are about to hit a production worker, they can quickly realize it and stop immediately.
To be aware of what’s around them, Vision Cobots leverage advanced computer vision algorithms and 3D vision-based cameras. Thanks to computer vision, cobots can gain a high-level perception and awareness from the world around them.
Computer Vision + Cobots: Vision Cobots
Cobots could work faster and be more productive. But the safety standards in the manufacturing industry enforce them to work at a lower capacity. Even though they are designed for secure collaboration with human workers, the smaller end-effectors and payload can still be harmful.
The right Artificial Intelligence (AI) and Machine Learning techniques can dramatically improve their safety because it can make them faster decision-makers.
Computer vision uses pattern recognition and learning techniques to make sense out of video images. It is the intelligence underneath the vision system of the cobot. With the right input from cameras and other sensors, these autonomous systems can see and be completely aware of the surroundings.
Veo Robotics, a popular vendor in the Industrial Automation market is developing a 3D sensing system that fulfills the requirements of the Speed and Separation Monitoring standard (SSM from ISO 10218 and ISO/TS 15066). This system uses 3D depth cameras and machine vision algorithms to give cobots full awareness of its surroundings and to be able to work with humans.
What are some of the benefits of the Vision Cobots:
- They create a safety factory automation environment by keeping track of abnormal and sudden external movements.
- They have a fast decision-making process for sudden stops, reduce speed, and change motion.
- They can be guided by the hand of a human for safe teaching and operation.
- They can also gather production-line operations and product quality data and send it to an analysis or management system.
Key Vision Cobots Components
To allow Cobots to work without the traditional safety fence they need to have some additional infrastructure that helps them see and process information better. By itself, a cobot has a variety of built-in sensors that help it be safer. Some of these sensors might be:
- Proximity sensors: These might help cobots know when to stop or slow down their operations to avoid bumping with their human co-workers.
- Force or torque-feedback: These sensors can help cobots improve their sensitivity.
- LiDAR: A system that measures the distance to an object using laser technology.
- Motion sensors: Detect movement near the working area.
- 2D-Vision sensors: Cobots have embedded 2D cameras that help them be aware of their environment.
But aside from these built-in sensors in the Cobot, the autonomous system needs more feedback from the external environment. They need flexible vision and powerful processing units to run the computer vision algorithms.
Vision and Compute
- 3D-vision based cameras. 3D cameras are essential elements in the cobots system. As opposed to the traditional 2D cameras which perceive flat images, these 3D devices provide the cobot with a human-like vision. In order to perceive the environment in a three-dimensional world, just like humans do, cobots need 3D vision.
- Computer-vision appliance: Advanced machine-vision algorithms can be quite compute-intensive. An industrial and embedded purpose-designed platform for computer-vision can be the key to high-quality vision outputs. An example of this platform is Lanner’s LEC-2137 which is optimized for computer-vision software. This platform runs powerful intel processor, HDMI and VGA ports, and high tolerance for temperature.
Cobots will not replace humans. They will only complement them in monotonous or high-precision tasks and work besides them. Cobots don’t have the super-human capacities that big industrial robots have. Their design is made to collaborate and operate outside safety fences. Cobots also give access to autonomy to a wide general public because they are much cheaper than big industrial robots, easier to install, and much safer.
In previous years, factories and labs have adopted cobots quite optimistically. These collaborative robots can monitor abrupt nearby movements or employees breaking into Cobot’s danger zone, and save humans from accidents.
Plus, the precision and human-like vision of these cobots are becoming quite extraordinary. Thanks to advanced machine-vision algorithms and its underlying custom-made computing power, a cobot can now process a massive amount of 3D video input and react quickly. It will detect obstructions near its work area, it will immediately shut down its movement to avoid any harm to human co-workers.