This project develops a new type of social intelligence for robot companions in assisted living environments. This new intelligence allows a robot to learn a person's daily activity and location without constantly following him or her. Such a human-aware capability frees the robot to do its daily routine work, while being able to attend the human more proactively and effectively. A large population can benefit from the outcome of this project, especially the elderly and patients who need assistance and companionship in their homes. In addition, people in their work places can also benefit from this project, since robot co-workers can provide routine service and emergency assistance with such human-awareness. The proposed educational activities will raise more awareness of robot research through curriculum enrichments in a series of undergraduate and graduate courses. Through the outreach activities, it will stimulate prospective and current college students to pursue degrees and careers in science and engineering.
The objective of this project is to develop the new theoretical/algorithmic framework and the open hardware/software platform for this type of considerate co-robot intelligence. The proposed method is based on ubiquitous human awareness-a robot capability of knowing a person's activity and location in an indoor environment without using onboard sensors. This capability is realized through wearable sensing and computing from a human-based perspective. The major research efforts consist of four parts: co-robot 3D semantic mapping; human activity and location inference; activity prediction and behavioral anomaly detection; experimental evaluation using open hardware/software platforms. A case study will be conducted to evaluate the effectiveness of the proposed considerate co-robot intelligence in elderly fall prevention, detection and intervention.
