rCommerce Laboratory
Human-Robot Teams
Robotics Institute
Overview
.
Prior Work
.
.
Year 1
.
Robotic solutions are addressing problems in increasingly complex domains in which a single agent acting alone cannot achieve the required goals. Disaster response, agriculture, mining, and construction all involve several spatially distributed tasks that must be executed by teams of autonomous agents such as robots or humans. The requirements of these and similar problem domains are driving the state-of-the-art in multi-robot systems and human-robot interaction.

As do humans, each type of robot or machine has unique capabilities that make it particularly suited to performing certain activities. In disaster response, a large automated excavator may be used to clear rubble from a collapsed structure, whereas a small robot with cameras and other sensors might be best suited to searching for victims in a collapsed building, and skilled human responders might be best suited to extricating survivors. Furthermore, although robots can perform many tasks, they have several limitations and may periodically require human intervention or assistance. For example, an excavator may be able to work autonomously 90% of the time, but require a human to take over operation when working in a location that requires particular care due to the presence of humans or underground power lines. There are also several tasks that are better suited for humans than for robots. Therefore, humans will not only be operators and perform interventions, but instead, will work as team members in human-robot teams that can have fluid hierarchies and role assignments. Thus, coordinating teams of humans and robots is an important problem for the future of robotics.

Some scenarios that might benefit from coordination are described next:
  • Following or during a natural disaster such as an earthquake or flood, multiple humans and robots can be deployed to assess and respond to the situation. Sub-teams of agents perform various categories of tasks such as monitoring, inspection, search and rescue, excavation, evacuation, and distribution of aid.
  • In agriculture, a number of automated haulage vehicles may need to rendezvous with automated harvesters in different locations in a field, to bring the harvested grain back to silos for storage.
  • In mining, activities such as cutting coal, creating roof supports, and transporting coal may take place simultaneously in various locations in a mine and be performed by a combination of humans and machines (including robots). The machines responsible for each of these activities must be efficiently coordinated.
  • In construction, activities such as excavation, earth-moving, transportation of building materials, and assembly take place simultaneously at different locations on the construction site. Again, the machines responsible for each of the activities must be efficiently coordinated. Some equipment may require oversight or intervention by a human during operation. As such, the human members of the team are part of the coordination effort.
  • Warehouse management involves moving products from one point to another in a warehouse in order to stock products or fulfill customer orders, sometimes requiring assembly. Robots can perform significant portions of this operation, but humans may be required to handle fragile items such as eggs and tomatoes. As such, robots need to be coordinated with each other as well as with human staff.
The specific characteristics of the coordination problem to be solved are highly dependent on the domain. In some cases, tasks are divided among agents who perform the assigned tasks with no interaction amongst themselves. On the other hand, some cases require constant and tightly-coupled interaction between members of a team. However, some underlying properties remain consistent, and a coordination mechanism that allows customization to specific domain needs and team (resource) constraints will be a powerful tool with great impact.