Carnegie Mellon University develops new algorithm for optimal assignment of work

Text |

Jack Ma once said that "artificial intelligence will completely change the mode of human employment". At present, artificial intelligence has become a new engine for future technological revolution and industrial transformation, and drives and promotes the transformation and upgrading of traditional industries. Artificial intelligence has a wide range of applications, from digital government to smart transportation, from industrial agriculture to financial education, to judicial medical care and retail services, and the impact of artificial intelligence on employment has become more and more obvious.

The prevailing view is that artificial intelligence can save humans from tedious, tedious work, giving people more time and energy to do something meaningful and creative. As robots are increasingly added to factory floors, warehouses and other settings, the need to optimize the distribution of work between humans and machines is becoming more and more urgent.

A new algorithmic planner developed by the Robotics Institute at Negie Mellon University achieves optimal task allocation between humans and robots. The algorithm, named "ADL," effectively answers the three questions of when a robot should take action to complete a task, when a robot should learn a new task, and when a robot should delegate tasks to humans.

Using algorithms and software to decide how to delegate and divide labor is not new, however, this work is one of the first to incorporate robotic learning into inference. Also, in manufacturing, where a worker controls a robotic arm, it takes time to teach the robot a task. If the algorithm can be successfully implemented, the work efficiency will be greatly improved.

To test the new algorithm, the researchers set up scenarios: humans and robots raced to insert blocks into pegboards and stack Lego blocks of different shapes and sizes. As a result, the new algorithms have shown good results. But in complex tasks, there is still room for improvement in the performance of robots.

In addition, "strong artificial intelligence" requires robots to do some work after predicting and learning new tasks. Based on this, the algorithm needs to transform the problem into a mixed-integer program—an optimization program often used for scheduling, production planning, or designing communication networks. In the future, it is hoped that through continuous updating and improvement, the algorithm can better serve human beings.