The latest study from the University of Massachusetts Amherst reveals that by programming robots to form their own teams and patiently wait for their teammates, tasks can be completed more quickly, offering potential benefits for manufacturing, agriculture, and warehouse automation. The research has been recognized as a finalist for the Best Paper Award on Multi-Robot Systems at the IEEE International Conference on Robotics and Automation 2024.
“There’s a long history of debate on whether we want to build a single, powerful humanoid robot that can do all the jobs, or we have a team of robots that can collaborate,” says one of the study authors, Hao Zhang, associate professor in the UMass Amherst Manning College of Information and Computer Sciences and director of the Human-Centered Robotics Lab.
In a manufacturing environment, a team of robots can be a cost-effective solution, maximizing the capabilities of each individual robot. However, the challenge lies in effectively coordinating a diverse set of robots with varying abilities and roles.
To address this challenge, Zhang and his team have developed a learning-based approach for scheduling robots known as learning for voluntary waiting and sub teaming (LVWS).
“Robots have big tasks, just like humans,” says Zhang. “For example, they have a large box that cannot be carried by a single robot. The scenario will need multiple robots to collaboratively work on that.”
The other behavior is voluntary waiting. “We want the robot to be able to actively wait because, if they just choose a greedy solution to always perform smaller tasks that are immediately available, sometimes the bigger task will never be executed,” Zhang explains.
In their evaluation of the LVWS approach, the researchers assigned 18 tasks to six robots in a computer simulation and compared their LVWS approach to four other methods. The simulation had a known, perfect solution for completing the scenario in the fastest time. They calculated the suboptimality of each method compared to the perfect solution, which ranged from 11.8% to 23% for the comparison methods.
The new LVWS method showed only 0.8% suboptimality, indicating its close proximity to the best possible solution. Williard Jose, a doctoral student in computer science at the Human-Centered Robotics Lab and an author of the paper, noted that the solution is close to the best theoretical solution.
Incorporating a wait time for robots can actually improve the efficiency of the entire team. Consider this scenario: You have three robots—two capable of lifting four pounds each and one capable of lifting 10 pounds. One of the smaller robots is currently occupied with another task, while there is a seven-pound box that needs to be relocated.
“Instead of that big robot performing that task, it would be more beneficial for the small robot to wait for the other small robot, and then they do that big task together because that bigger robot’s resource is better suited to do a different large task,” says Jose.
“The issue with using that exact solution is that computing it takes a really long time,” explains Jose. “With larger numbers of robots and tasks, it’s exponential. You can’t get the optimal solution in a reasonable amount of time.”
In their analysis of models involving 100 tasks, where calculating an exact solution is infeasible, they discovered that their approach enabled the completion of tasks in 22-time steps, compared to 23.05 to 25.85-time steps with the comparison models.
Zhang envisions that this research will advance the capabilities of automated robot teams, especially in scenarios involving scalability. For example, he suggests that a single humanoid robot might be more suitable for the limited space of a single-family home, while multi-robot systems are more suitable for large industrial settings that require specialized tasks.
Journal reference:
- Williard Joshua Jose, Hao Zhang. Learning for Dynamic Subteaming and Voluntary Waiting in Heterogeneous Multi-Robot Collaborative Scheduling. 2024 IEEE International Conference on Robotics and Automation (ICRA), 2024; DOI: 10.1109/ICRA57147.2024.10610342