Development 

            A rumble strip, used as a safety device on freeways to alert drivers that they are veering off the road, inspired the Rumble Sensor. We realized that we can calculate the speed of a vehicle from the sound frequency produced by a vehicle driving over the rumble strip. However, because of the inconvenience of laying down a rumble strip on the curved testing track and wherever else we wanted to roll our vehicles, we reimagined the idea. Instead, we fabricated spokes that we inserted into one of the four wheels on the roll car and placed a card that raked the spokes to produce regular interval sound when the vehicle rolled. 

 

Implementation

            We programmed a Raspberry Pi and microphone to record the rumbling sound, which was now more of a clicking sound. We analyzed the recorded sound using Fast Fourier Transform (FFT) on MATLAB. For speed comparison, we installed quadrature encoders on the two front axles of the two roll cars and used python to program the encoders. The encoders, however, skipped pulse counts. We decided to pivot and use slow-motion video to capture the path of the roll car, analyzing the distance and time traveled by the car along the path and comparing it to the speed we found using the rumble sensor. We performed multiple runs with a single roll car, recording its speed and multiple runs with two roll cars, with the rumble sensor car recording its speed and that of the secondary car. We also tested using an encoder to measure the speed of the roll car, but we found that the processor of the Raspberry Pi lagged. 

 

Conclusion

            We were able to successfully calculate the speed of a single roll car with a microphone using both findpeaks and FFT analysis. The analysis provided higher resolution speed data than was possible with the slow-motion video. However, high levels of noise in our testing setup prevented us from achieving our goal of measuring the speed of two roll cars simultaneously with a single microphone. This objective is likely achievable if we are able to significantly reduce background noise. 

​