Postal Sanity

Michael Ravnitzky, chief legal counsel to US Postal Regulatory Commission (PRC) chairwoman Ruth Goldway, presented the following paper in June 2010:  “Offering Sensor Network Services Using the Postal Delivery Vehicle Fleet”.

Today,  ePostal News has commented on Mr. Ravnitzky’s paper in a rather entertaining fashion and pointed out the lack of connection to the postal core business.  In fact, the ePostal News article humored us enough to revisit Mr. Ravnitzky’s paper and re-evaluate the technical feasibility of these mobile sensor networks.  Unlike  Mr. Ravnitzky, we foresee a host of technical problems, however, there are a few applications which could promise some merit. Long story short: We edited a table in Mr. Ravnitzky’s paper, adding our opinion about the technical feasibility of these mobile sensor applications. (see the most right column, and yes, we left some blanks).

TABLE  1:   Matrix  of  Potential  Applications  for  Postal  Truck‐Borne  Mobile  Sensors

(1)   Sensors for chemical and biological agents often do not deliver real time results, depending on the agents to be detected. Often these sensors rely on accumulating agents in a filter media first. After collection these filters are then automatically examined in a costly stationary diagnostic unit. These means that any diagnostic results will likely be available with a delay of 10 or more hours. Further it means that any positive detection can only be assigned to a postal route, rather than a specific location. Given the severity of such agents and the lack of real time response of this approach we evaluate the technical feasibility as slim.

(2)   Detecting radiological materials yields real time results, which can be geo-mapped. Assuming the availability of low cost sensors this should be possible.

(3)   Air quality is a concern in urban areas. However a postal truck sitting in a LA traffic jam likely measures elevated values, not representative of residential areas. Data needs to be filtered for meaningful results. Another problem is the lack of 24/7 coverage, which would be preferable for this kind of data. Stationary units do not only provide data around the clock, but that data can be more meaningful since it relates to a single fixed location.

(4)   This should be possible assuming the availability of low cost sensors, able to detect trace amounts within a brief time exposure (since postal truck keeps on moving). However, shifting winds may lead to fake data. Fake data will also be produced by temporary human induced emissions

(5)   See (3)

(6)   Acceleration sensors work only if the driver does not avoid the potholes, like most drivers do. A (not so serious) brute force approach may consist in instructing the drivers to hit every pothole at full speed. Optical sensors may be a possibility as well, but they have their own pitfalls.  A very low tech approach could consist in the driver hitting a button as he passes by potholes, alas this approach is subjective and not very scientific.

(7)   Cars are moving objects. A license plate scanning OCR should thus have a real-time link to law enforcement, sending plate number and location data.  Expect legal and privacy issues.

(8)   Meaningful noise profiling is never based on single samples. Noise profiling involves long term measurement at a single location to detect peaks and calculate noise averages

(9)   Expect legal and privacy issues. The videos unlikely give the history of interesting events at a location since the postal truck keeps on moving.  Additionally people are moving objects as well.