A Low-Power Motion Capture System with Integrated Accelerometers Riccardo Barbieri, Elisabetta Farella, Luca Benini, Bruno Ricc´ o DEIS - University of Bologna 40136 Bologna, ITALY Email: rbarbieri,efarella,lbenini,bricco@deis.unibo.it Andrea Acquaviva ISTI - University of Urbino 61029 Urbino, ITALY Email: acquaviva@sti.uniurb.it Abstract— Motion capture is an emerging technology enabling the design of natural user interfaces for wearable devices based on gestural recognition. However, costs and energy requirements are critical factors to enable their diffusion to low-end wearable systems. Current commercial products do not match these requirements. For this reason, we developed a low-cost/low- power wearable motion tracking system based on integrated accelerometers called MOCA (Motion Capture with Accelerom- eters). Our system is composed by sensing units connected to a control/acquisition board responsible for reading and pre- processing data and a mobile terminal running the recognition algorithm. Experiments performed to validate accuracy, power consumption and real-time performance demonstrate low-power and flexibility features of the proposed tracking system as well as its effectiveness as input interface. I. I NTRODUCTION Motion tracking is a key enabling technology for research and commercial fields ranging from Human-Computer Inter- action to Robot Navigation, Virtual and Augmented Reality. Tracking is defined as the process of tracing the coordinates of moving objects in real time using a set of sensors whose number and characteristics depend on the number of degrees of freedom (DoF) to be monitored and on the target application. Tracking systems are used in ambient intelligence applica- tions to provide users with personalized services depending on their location or position. They are also used in augmented reality applications to generate virtual scenes depending on the user point of view. Present and past research on motion tracking mainly concerns the design of high accurate and expensive sensor systems. Simpler tracking tasks, such as ges- tural recognition do not require high precision, but currently available commercial systems based on electromagnetic and optical sensors are too expensive for consumer products. Our work addresses the problem of finding a low-cost/low-power solution based on commodity components to enable a simple but effective gestural interface. In this paper we propose a system solution based on low- cost and low-power accelerometers to track human motion as input interface to the machine. The system is called MOCA (MOtion Capture with Accelerometers), a motion capture system based on integrated accelerometers and a wearable computer for sensor data processing. The system is com- posed by the accelerometer units connected to an acquisition control board that communicates to the wearable computer through the standard serial port. The wearable device runs the acquisition/pre-processing software and the monitoring algorithm. However, pre-processed data can be also sent to a remote sensor fusion system for ambient intelligence and location awareness applications, thanks to the wireless con- nectivity of the device. The use of a general purpose palmtop computer as a pre-processing stage provides high flexibility and reduces the need for additional dedicated hardware. The system is completely based on commodity components. Cost, excluding the palmtop computer, is lower than $100. The proposed solution is flexible, as the acceleration sensors can be placed anywhere on the body. This is in contrast with system targeting specific body parts, such as the well known glove applications [7], [13]. Moreover, the use of integrated accelerometers and a palmtop computer as processing platform ensures flexibility and low power consumption, as well as unconstrained mobility. We performed a number of tests to verify accuracy, power consumption, real-time performance. Tests performed on the prototype shown 3 degree accuracy for different contiguous positions tracked. As regards dynamic characteristics, the system provides a correct measure of typical human body movements (up to 3 radians/sec). Power consumption mea- surements show that the system can operate continuously at least for 12 hours with 100g, 750mAh batteries. The rest of the paper is organized as follows. In Section II we give an overview of previous work in this area. In Sec- tion III we describe the system architecture. Finally, we show experimental results in Section IV. Section V concludes the paper. II. RELATED WORK Robot navigation and vision systems need fast and accurate sensors characterized by reduced weight, small size and low power consumption, leading to high cost systems. For this reason, inertial sensors (micro-machined accelerometers and gyroscopes) that are fast and robust but with limited accuracy are coupled with non-inertial tracking systems based on GPS or vision [1], [8], [11]. Augmented Reality (AR) applications also require high accuracy, together with low latency and jitter. Their purpose is to track the human motion in order to produce virtual scenes enhancing the actual view of the user in real time. Being 0-7803-8145-9/04/$17.00 ©2004 IEEE. 418