About this project
One of the major challenges in robotics is to have mobile robots operate in environments that are inhabited by humans, such as homes, shops, warehouses, and hospitals. Robots in such settings can potentially have great economical benefits and can increase the quality of life of people. However, for robots to be suitable for more than the simplest and most predictable tasks, they need to
take the activities of the human inhabitants into account. For instance, it might be better for the robot to plan to clean the living room while the human is taking the dog for a walk, than while the human watches TV. The robot may also plan to clean the kitchen after the human has been preparing lunch, and so on. Our project will develop methods for robots to plan in a manner that takes human activities into account. These methods will consist of two major steps:
(1) Observe the humans, understand what they are doing and predict future activities based on this. To observe the human the robot uses various sensors (video cameras, motion detectors etc) distributed in the environment, and uses algorithms for activity recognition and information about the environment (e.g. which room is the kitchen) to understand what is going on.
(2) Find a task plan for the robot which takes the predicted future activities of the humans into account. In doing so, the robot needs to take into account the effects of both its own actions, and those of the humans.
On grant from Vetenskapsrådet, Swedish Research Council (about 1,600,000 SEK).