Robotic computer algorithm to revolutionise task delegation

A research team from the University of Carnegie Mellon (CM) have developed a robotic computer algorithm to simplify the task delegation process.

How typical is robotic assistance in the workplace?

It is becoming more and more common to employ robotics to improve work efficiency, such as in factories and warehouses. However, it has been noted that robots are not necessarily appropriate for all jobs, and some instances require human interference. Thus, scientists have developed a robotic computer algorithm that successfully delegates tasks between robots and humans, depending on complexity and importance.

This research was funded by the Office of Naval Research and the Army Research Laboratory.

How does the robotic computer algorithm work?

Scientists from CM’s Robotics Institute (RI) have introduced an algorithmic planner known as: ‘Act, Delegate or Learn’ (ADL). As stated, the robotic computer algorithm assigns tasks based on necessity; in other words, whether the robot needs to ‘act’ and take the task, ‘delegate’ the assignment to a human or ‘learn’ how to complete the job.

The researchers asked three questions to develop this technology successfully: When should a robot act to complete a task? When should a task be delegated to a human? And when should a robot learn a new task?

“There are costs associated with the decisions made, such as the time it takes a human to complete a task or teach a robot to complete a task, and the cost of a robot failing at a task,” explained Shivam Vats, lead researcher, and a PhD student at the RI. “Given all those costs, our system will give you the optimal division of labour.”

Researchers determined this technology to be successful by testing the computer algorithm in scenarios where robots and humans had to insert blocks into a pegboard and stack parts of different shapes and sizes.

How helpful is this technology?

The introduction of this robotic computer algorithm has the potential to help manufacturing and assembly points, in sorting packages, and in any environment where humans and robots can collaborate to complete several tasks.

Utilising algorithmic software to decide and delegate jobs is not revolutionary, even when robots are included in the team. However, this robotic computer algorithm is the first to consider learning in its reasoning.

“Robots are not static anymore,” Vats said. “They can be improved, and they can be taught.”

The costs and drawbacks of employing robotic technology

Robotic arms utilised in manufacturing are manipulated by humans to educate the robot. Thus, it takes time and a high upfront cost for the robot to learn how to complete the task appropriately.

However, the long-term benefits of teaching a robot a new skill arguably outweigh the drawbacks. Scientists noted that part of the complexity involved is deciding when is best to teach the robot versus delegating the task to a human—this requires predicting what other jobs the robot can complete after learning the new task.

How does the computer algorithm divert this issue?

Researchers have developed a mixed-integer computer program to prevent this from becoming a problem. This optimisation program – commonly employed in scheduling, production planning, and designing communication networks – can be solved efficiently by off-the-shelf software. Researchers noted that the planner performed better than traditional models in all instances and decreased the cost of completing the tasks by 10% to 15%.

This work was presented at the International Conference on Robotics and Automation in Philadelphia. The research team included Oliver Kroemer, Assistant Professor at the RI, and Maxim Likhachev, Associate Professor at the RI.

Subscribe to our newsletter


Please enter your comment!
Please enter your name here

Featured Topics

Partner News


Latest eBooks

Latest Partners

Similar Articles

More from Innovation News Network