Work

Wearable Computing

Year: 2011
Type: article
Abstract: In robotics, activity-recognition systems can be used to label large robot-generated activity data sets. It enables activity-aware human-robot interactions (HRIs). It also opens ways to self-learning autonomous robots. The recognition of human activities from body-worn sensors is a key paradigm in wearable computing. In this field, the variability in human activities, sensor deployment characteristics, and application domains has led to the development of best practices and methods to enhance the robustness of activity-recognition systems. We argue that these methods can benefit many robotics use cases. We review the activity-recognition principles followed in the wearable computing community and the methods recently proposed to improve their robustness. These approaches aim at the seamless sharing of activity-recognition systems across platforms and application domains. Finally, we outline the current challenges in wearable activity recognition. Recognizing, sharing, and reusing robot behaviors across multiple robot platforms with various similarities are challenging. Although descriptions for objects [e.g., computer-aided design (CAD) models and recognition models] and environments (e.g., geocoordinates, local coordinates, and feature maps) are largely interchangeable across different robot hardwares, robot task descriptions are typically hardware dependent. This has prevented the generation of generic data sets for robot behaviors. However, such data sets are important and underpin many of the algorithmic advances in object recognition [1]-[3] or in the creation of joint world models [4]-[7]. It has also hindered the progress of robot cognition and robot learning by preventing the robots to understand and learn from each other's actions. © 2011 IEEE.
Language: English
Open Access status: green
OpenAlex ID: W1824200880
FWCI: 8.792
Cites: 70
Cited by: 44
Related to: 10