The system design detailed in the previous section enabled us to realize the desired results from AroundSound. Overall, we were able to use the system to generate the 3D audio from the distance measured by the ultrasonic sensor, which was measured over a range of angular points driven by the servo. In this section, we outline the results we observed from our design below, as well as some of the test results that drove our decision process. We also detail the safety and usability considerations associated with this device, and how it may impact the general community.
From our tests and demonstrations, we found that AroundSound’s performance was extremely reliable. We set up multiple test environments consisting of movable segments of cardboard in addition to the obstacles (walls, cabinets, etc.) present in the lab environment, and observed that AroundSound produced notes of differing frequencies depending on the distance of an obstacle from the system. In particular, we noticed that the proximity of the surface played a significant role in how many times it was detected over the angular sweep and whether there were any surfaces obstructing the ultrasonic sensor’s path - this would be particularly useful for navigating around large floor obstructions and walls.
Programming the angular sweep to take six seconds from start to finish caused there to be some slight dissonance between our expectations regarding the audio sweep and the actual sound - it proved a little more difficult to discern the intensity and time difference between the directional components of the sound at first, but with some focus it became much simpler to identify the location of the obstacle. We did try reducing this time interval, but found that the reduced interval did not allow for fully accurate distance measurement and processing.
Our demonstration with Professor Adams showed that it was simple for a user who had never used the product before to detect the location of obstacles around us. It took only two executions before he was able to identify the proximity of obstacles at angular positions around him, proving that the device does its due diligence to its intended purpose. We had all tried testing the device prior to this demonstration with multiple obstacles of different sizes, and found a similar correspondence before being able to identify the positions of obstacles around us.
We did investigate some alternative methods for our subsystems while developing AroundSound, particularly with respect to the sound and the sensor subsystems.
Our initial 3D audio design was inspired by a simpler series of calculations that factored in the distance between ears and the angular orientation of the source of the sound in calculating the interaural time difference. However, this approach was soon discarded in favor of a well researched understanding of the ITD and IID components of 3D audio generation. After completing this implementation, we found that direct audio from the PIC32 sounded extremely grating on the ears, and even amplifying the audio through the gain circuitry of the speakers did not produce an appealing sound. Finally, we chose to use FM synthesis with discrete note mappings to generate clean tones with an understandable result of the distance from the system.
We tried to use different sensors as well - in particular, we had hoped to use either a laser time of flight sensor or an analog infrared distance sensor for distance measurement, but we found that these sensors either posed a significant problem while attempting to read data or were extremely inaccurate with their distance measurements. We then chose to work with the ultrasonic sensor, which provided surprisingly accurate results and was simple to set up on the PIC32.
This prototype of AroundSound has long cables that connect the servo gimbal and ultrasonic sensor to the PIC32. These wires could potentially get entangled with some obstacle and pose a serious hazard to the user, even though the current prototype is not very mobile. Future iterations (if any) of the product should factor this hazard in and attempt to have a single board design that minimizes this risk. There are no other significant safety concerns regarding this product.
Our demonstrations have shown that the AroundSound does satisfy its intended purpose - however, it should be noted that the device performs well only when considering wide (in the tens of centimeters range) objects with hard surfaces. The sensor at present cannot detect obstacles that are soft, narrow, or below a meter in height, imposing some restrictions on its use cases. By incorporating a more robust set of sensors across the AroundSound, it may be possible to overcome these limitations.