Computer Vision and Robotics Use Case

THE CHALLENGE
The first industrial robot was invented in 1954 and was installed seven years later in a General Motors factory for spot welding and die-casting. Since then, robotic technology has been used in industries from manufacturing to agricultural farming as a means to increase efficiencies, lower costs, and increase revenues. These robots are usually designed to work independently, executing pre-scripted tasks in spaces protected from human interference. They have increased factory productivity but have limited capabilities.

Cobots, or collaborative robots, are a new step in industrial robot technology. Unlike most robots, which act as replacements for human workers (and often operate in cages to prevent injury to workers), cobots are designed to work sideby- side with their human counterparts, even collaborating on the same task. How do these robots gain new abilities that can increase their operational value while remaining safe and secure as they operate in a factory near humans?

THE APPROACH
One way to increase robotic abilities in a safe and efficient manner is to use an innovative new technology: computer vision. This technology enables a robot, or computer system, to use a camera or scanner to transform multidimensional inputs into data it can process, “perceiving” its surroundings and mimicking sight. Computer vision coupled with machine learning gives the computer increased technical abilities and the opportunity to perform more complex tasks. Robots accessing computer vision gain abilities beyond scripted tasks and can augment the abilities of their human coworkers by participating in their labors or by using technologies such as infrared imaging to see and report on things invisible to the human eye.

This technology dramatically increases the potential for robotics in industry, creating avenues that would not otherwise be viable. For example, using an AI-enabled cloud, connected robots could recognize objects faster and send collective messages, notifying or warning humans of situations that they could not see. They could also aid in quality control, as they could be able to recognize the condition of products when compared against the expected visual representation.

Similar advantages could pertain to agricultural production. Independent robots using computer vision could differentiate between product quality levels; for example, the robot could use imaging types in both visible and ultraviolet light to detect below-surface discrepancies and extract a higher profit from varying products by identifying food grades. It could even warn for diseases, such as peach leaf curl on trees, that would significantly reduce productivity if not treated.

THE RESULT
Currently, industrial robots harbor many potential safety dangers, as they have no awareness of their surroundings other than what is provided by sensors; this could cause serious harm to people working nearby or alongside them. However, with the addition of new sensing technologies, robots could be used in closer proximity to humans and in more confined spaces, so that factory workers and robots would be able to safely work in tandem. Both the production capacity and the safety of the factory could increase. Robots could perform more complex tasks, and they could operate in a disordered space by recognizing the objects they should interact with.

Wind River offers solutions that incorporate the latest ROS 2 framework, so developers can focus on application development, leading to more innovative robotics. Compute and partitioning capabilities can protect safety applications while providing the high performance that is important to enhance further collaboration between humans and robots.

To learn more contact our sales inquiry desk.