Towards a Minimal Architecture for a Printable, Modular, and Robust Sensing Skin
Abstract
Here we present an adaptation of sensitive robotic skin to the domain of rapidly prototyped systems, with the long-term goal of integrating sensing and networking with the manufacturing process using low-cost printed flexible-substrate semiconductors. In addition, this skin will inform the robot as to the shape and arrangement of the sensors on the surface, as well as recovering from and reporting changes in this topology if damaged. By exploring this approach to whole-body multi-modal sensing, we expect to improve the ability to localize and navigate in unstructured environments via detection of obstacles, targets, and environmental conditions. The shape sensing feature will allow the robots to modify their behavior and continue functioning as long as possible despite destructive modifications of the robot. For example, a legged robot with such a skin integrated throughout its body will be able to detect a leg loss, and modify its gait accordingly to continue locomotion. As a first step, we consider how the design constraints of such a sensing skin motivate the design parameters of a sensor network, and explore a particular network architecture that achieves the desired data integrity with simple underlying logic. Ultimately we want to answer the question: what are the minimum necessary conditions for constructing a serviceable, highly configurable sensor skin that is compatible with low-performance printed electronics?
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 27, 2014
- Accession Number
- ADA605322
Entities
People
- Austin Buchan
- Jonathan Bachrach
- Ronald S Fearing
Organizations
- University of California, Berkeley