The development of a bio-inspired vibration source localization algorithm

Vibration source localization is an interdisciplinary topic that has applications in both engineering and biology. To develop bio-inspired technologies that can be applied in engineering, a vibration source localization algorithm is developed based on a spider dynamic model. The model includes: (i) a point source; (ii) a spider model with 8 legs resting on a non-dispersive substrate that introduces correlated leg inputs. The proposed algorithm first calculates the leg inputs using a backward filter forward smoother algorithm with given leg responses to the excitation from the vibration source; and then estimates the source angle, source distance, wave speed and decay rate through optimization. The optimization utilizes the time and amplitude relationships between the leg inputs and the leg-substrate contact locations. The effectiveness of the algorithm is validated by comparing the algorithm estimations with the actual parameters. Results show that the algorithm gives accurate estimation of the source angle and wave speed usually with errors within 3 deg and 1 m/s, respectively, but cannot estimate the source distance. Two spatially-separated spider-like models are needed for estimating the source distance. The algorithm works well even under low signal-to-noise ratio down to 11 dB.