Orientation tracking for humans and robots using inertial sensors

Open Access
Publisher Website
Google Scholar
Abstract
Joint angle determination for robots with flexible links can be difficult. Inertial orientation tracking combined with RF positioning provides an accurate method for determining end effector orientation and location. The same technology could also be used to determine human posture for the purpose of inserting humans in synthetic environments. Orientation filters based upon Euler angles suffer from singularities. This paper describes the design, implementation, and preliminary testing of an inertial tracking system using a "complementary " filter based upon quaternions. This filter is capable of tracking a rigid body through all orientations and is more efficient than those based on Euler angles. Results of qualitative tests of a prototype inertial angle tracking device are presented. Background Precise control of a robot manipulator requires accurate determination of end effector location and ori- entation. Due to link flexing and bending, this is often difficult. This problem could be solved using a hybrid inertial tracking technology currently being developed at the Naval Postgraduate School for the purpose of inserting humans into a networked synthetic environ- ment (SE). The location and orientation of an end effec- tor could be ascertained in the same way as the body posture of a tracked human. Just as the body posture of the tracked human could be used to precisely control the visual and audio queues of given to a user of a SE, so too could the orientation and location of an end effector be used to precisely control a robot manipulator. Tracking systems currently in widespread use include optical, magnetic, and acoustic systems (19). These systems typically have fairly high latency, mar- ginal accuracy, moderate noise levels, shadowing prob- lems, and limited range. The primary reason for most of these problems is the dependence of these systems on a transmitted "source" to determine orientation and loca- tion information. This source may be transmitted by body based beacons or received by body based sensors. Either way, limited range, shadowing problems and sus- ceptibility to interference makes such sourced systems unfit for tracking multiple users in a large working vol- ume. The largely "sourceless" nature of inertial orienta- tion tracking makes possible a full body tracking system that avoids the problems associated with current tech- nologies. Inertial orientation tracking is based upon the same methods and algorithms as those used for missiles, aircraft, and ships. Inertial angle tracking involves plac- ing multiple Magnetic Angular Rate Gravity (MARG) sensor units on the human body. Each MARG sensor contains three orthogonal rate sensors, three orthogonal accelerometers and three orthogonal magnetometers. Integration of angular rate sensor data provides the information necessary to calculate an orientation for each human body segment. Drift and scale factor correc- tions to these orientations are made continuously based upon accelerometer and magnetometer inputs. Body posture is determined based entirely on limb orientation and length. Complex kinematic routines are not neces- sary (9). The inherent noise. manufacturing defects, and

This publication has 6 references indexed in Scilit: