Human gait is very individual and it may serve as biometric to identify people in camera recordings. Comparable results can be achieved while using the acoustic signature of human footstep sounds. This acoustic solution offers the opportunity of less installation space and the use of cost-efficient microphones when compared to visual system. In this paper, a method for person identification based on footstep sounds is proposed. First, step sounds are isolated from microphone recordings and separated into 500 ms samples. The samples are transformed with a sliding window into mel-frequency cepstral coefficients (MFCC). The result is represented as an image that serves as input to a convolutional neural network (CNN). The dataset for training and validating the CNN is recorded with five subjects in the acoustic lab of DLR. These experiments identify a total number of 1125 steps. The validation of the CNN reveals a minimum -score of 0.94 for all five classes and an accuracy of 0.98. The Grad-CAM method is applied to visualize the background of its decision in order to verify the functionality of the proposed CNN. Subsequently, two challenges for practical implementations, noise and different footwear, are discussed using experimental data.
14 September 2005 in Proceedings of the 37th International Symposium on Automation and Robotics in Construction (isarc)
In the past years parallel robots demonstrated their capability in applications with high-dynamic trajectories. Smart-structures offer the potential to further increase the productivity of parallel robots by reducing disturbing vibrations caused by high dynamic loads effectively. To investigate parallel robots and their applications, including suitable control concepts for smart-structures, the Collaborative Research Center 562 was founded by the German Research Council (DFG). The latest prototype within this research center is called Triglide. It is a four degree of freedom (DOF) robot with three translational and one rotational DOF. It realizes an acceleration of 10 g* at the effector. In the structure of the robot six active rods and a tri-axial accelerometer are integrated to control effector vibrations in three translational DOF. The main challenge of this control application is the position dependent vibration behavior. A single robust controller is not able to gain satisfying performance within the entire workspace. Therefore a strategy for describing the vibration behavior by linearization at several operating points is developed. Behavior in-between is approximated by a linear approach. On a trajectory robust controllers in all operating points are smoothly switched by robust gain-scheduling. The scheduling parameters are fast varying and though a suitable stability proof is defined, based on Small-Gain approach. Several transformations enhance the results from Small-Gain Theorem and reduce the usual conservatism. Experimental data is used to show the improvements made.
3 July 2013 in Journal of Intelligent Material Systems and Structures
For next generation aircraft, contra rotating open rotor propulsion systems are currently discussed. Their economic advantages are in conflict with their high noise emission, which concentrates in distinct frequency bands. To bring contra rotating open rotor engines into operation at commercial aircraft and to maintain the passenger comfort level, active systems for noise reduction are considered. In this article, a contra rotating open rotor noise simulator for the future testing of active systems is presented. The simulator consists of a 14 × 8 loudspeaker array, which is placed close to a test fuselage. Driving each loudspeaker by a particular signal, complex sound pressure fields can be synthesized on the fuselage. The algorithm presented here calculates loudspeaker signals to synthesize the target sound pressure spectra on the fuselage. These target spectra were derived from numerical contra rotating open rotor engine simulations, coupled with a Ffowcs Williams–Hawkings solver to propagate the calculated sound pressure field toward virtual surface microphone positions on the test fuselage. Requirement for such synthesis is the knowledge of the transfer paths from all loudspeakers to all surface microphones. In experiments, these paths are measured, and the loudspeaker signals are synthesized by the algorithm. Finally, measurements with the loudspeaker array and a microphone array are shown which prove the concept.