A Modular Approach to Sensor
Intelligent Systems and Robotics Center, Sandia National Laboratories, Albuquerque, New Mexico
In this chapter we will demonstrate a method whereby sensor information can be integrated
into a robot-telerobot system in a modular fashion. The focus will be on constraint
information imposed by proximity sensors such as capacitive and ultrasonic sensors, but the
method can be extended to vision or force sensor feedback.
The framework is a nonlinear control system called SMART (Sandia's Modular Architec-
ture for Robotics and Teleoperation) that utilizes passivity and network concepts [1-3].
Each module in the SMART system represents an input device, a robot, a constraint, a
kinematic mapping, or a sensor input and can be represented by a network equivalent.
Systems are derived by combining modules in different telerobotic behavioral modes.
SMART has been described in various other publications [4-7], and has been applied to
problems as diverse as multirobot control, redundant robot control, and waste storage tank
operations. In this chapter, however, we will focus on sensor integration issues. In so doing
we will delve into the theoretical specifics of SMART, demonstrating how modularity is
achieved in a discrete domain and how nonlinear mappings can be integrated with sensor
data to enforce constraints. The system we will develop will be modular, both theoretically
and in actual implementations. Any module in a SMART system can be run on any
processing unit in a multiprocessor environment.
Modularity is especially important when integrating multiple proximity sensors in
telerobotic systems, because there are various potential sensors with complementary capabil-
ities, which can be configured in numerous different ways. For instance, capacitive sensors
such as the Sandia WHAP (Whole Arm Proximity) sensor are good for short-range (4 to 12
inches) detection of metals and other materials with a distinct dielectric constant. They can
see these objects instantaneously without reflections and independently of surface properties.
Ultrasonic sensors, on the other hand, operate independently of object materials but are
sensitive to surface textures and have an inherent lag in reading distances. Infrared sensors

Get Control in Robotics and Automation now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.