A wide range of businesses are looking to shorten production cycles, including those in the automotive, manufacturing, chemical, pharmaceutical and food industries. A common aim is to deliver cost-efficiencies and bring products to market quicker, while ensuring this is done with respect to workers’ rights and needs, as outlined in the Industry 5.0 framework. AI-enabled collaborative robots are one way to achieve this. “Robots capable of working in a collaborative way with humans have even more potential to increase production efficiencies,” notes Zoe Doulgeri, a professor from the Automation and Robotics Laboratory at Aristotle University of Thessaloniki in Greece, and project coordinator of CoLLaboratE (Co-production CeLL performing Human-Robot Collaborative AssEmbly). She continues: “Consider a situation where a human worker is able to demonstrate to a robot how to assemble new parts in just a few minutes. The robot is able to learn from the demonstration and adapt to changes in the environment so that it can really assist workers in their daily tasks. Our vision is of a future where humans and machines can collaborate effectively and flexibly in a shared workspace.”
The key objective of the EU-funded CoLLaboratE project was to develop industrial robots that can not only learn from humans, but also work alongside them safely. “We wanted to ensure that non-experts would be able to teach robots assembly tasks,” Doulgeri explains. “To achieve this, we developed different ways of teaching. These include visual demonstration by observing the worker performing the assembly, physical guidance (i.e. the worker takes the robot by the hand and leads it through the task), and augmented reality via a mobile app.” In terms of human safety, the project team developed software that ensures AI machines are fully aware of the presence of human co-workers on the production line, avoiding collisions while staying compliant, adaptive and accurate. “For example, we developed a new robotic skin that not only detects contact with objects, but can discriminate the type of contact,” explains Doulgeri. “Using deep learning, the robot can detect voluntary contacts of the human from involuntary and react appropriately in a safe manner.” With the robot having learned the task from demonstration, the project team created methods to bring AI and adaptive control into play so that the robot can autonomously improve itself and adapt to different scenarios. For example, the robot can share the load with a worker during collaborative handling of different objects. The robots are able to recognise the gestures of workers, and translate these into actions. In addition to these technical developments, the project team also paid attention to the social aspects of having AI robots on the production line. “The acceptance of robots by workers is of key importance,” says Doulgeri. “We therefore worked to ensure good communication between robots and humans, in order to build trust.”
Wide industry potential
Doulgeri expects that the advances made during the CoLLaboratE project will eventually lead to the increased adoption of collaborative robots in industry. She points out that in addition to the straightforward gains of increased productivity, working conditions of employees can be improved. More physically demanding and repetitive tasks can be delegated to robotic partners within the envisioned human-robot workspace. Potential end users of CoLLaboratE technology include automotive, aircraft and home appliance manufacturers. Many industries have already identified the potential of collaborative robotics, and are taking steps toward integrating the results of the CoLLaboratE project in their assembly lines. “We aim to not only access big industries, but also provide assistance to SMEs that have small-batch manufacturing,” adds Doulgeri. “They can also benefit from the flexibility offered by CoLLaboratE.” To finance the commercialisation of the project’s research results and scale up production, the project team is considering the possibility of a joint venture. Private and public funding is currently being sought. “Through this model, further development and testing can be conducted in an operational environment,” says Doulgeri.
CoLLaboratE, AI, robots, industrial, productivity, workspace, workers, production