Researching Sonar Sensors

The Too Smart City Sign,  in its current,  and likely final interactive structure, requires that it knows when users are standing near it,  and at what direction and distance from the sign structure.

Key challenges include:

• Users having variable height
• More than one user being at the location
• Non-user objects
• Extending range to entire 360 degrees  around sign
• Determining distance to a high precision.

First was to figure out what type of sensor would work.  I'd used sonar sensors before, and know that IR sensors are used in security systems for very cheap.  There are ofcourse binary solutions, like lasers or light beams that detect when an obstruction is present.

I found this very helpful post at Society of Robots on IR vs Ultrasonic.  As I'd suspected, sunlight will have too large of an impact in an outdoors park to allow IR to be a possibility for the SIGN.  It should be a good solution for the trash can though, since it will contained within the dark recesses of the bin.

First Idea:

In this version there would be a series of 8 sensors that are embedded in the mid-section of the body.Each sensor would tell user distance, and the sensor ID would be used to determine which direction the user was in.

Key problem with this is that it is overkill if we are using sonar sensors. Sonar sensors read with a beam, where the sensor area is first thin, then grows in width to the extension of its range.

This beam shape enables us to have an overlap between two sensors, therefore giving us twice as many directions, and halving the required number of sensors. The below image is a rough depiction:

The concern with this system is that if there are two users standing equidistant from the pole in two of the locations (as depicted by the orange dots in 5 and 7) it would be the same as if a single user were standing in zone 6.

Four sensors are still rather pricey; the sensors we are currently considering are MaxBotix and are $100 each.

SO ... perhaps a better idea is to use a single sensor and a stepper motor with a narrow field. The motor would turn around as fast as it can while still registering accurate sonar readings at each point.  There are some big problems with this approach.  The motor rotating constantly for 3 months outside may have problems, at least with continually providing accurate position.  Secondly, the cables connecting to the sonars sensor will need to handle turning repeatedly back and forth as the motor rotates 360 degrees in alternating directions.

This picture roughly depicts the idea - where it can determine different distances at different times.  This would then need some algorithm to analyze fluxtuations in distance over time.