The system tested was a direct proximity RF signpost system utilizing overlapping signposts operating at 49.86 MHz. Testing was divided into three categories, (1) random-route tests (taxis, police, pick-up and delivery, dial-a-ride), (2) fixed-route tests for transit-type vehicles, and (3) special tests for verifying performance under any conceivable urban environmental condition. The location system was specified to provide a basic vehicle location accuracy of at least 3000 feet at 95 percent and 450 feet at 99.5 percent of all test locations. At each checkpoint the location of the vehicle, as determined by the AVM System, was compared to the vehicle's actual location, the radial difference between these locations being the location error. For random route tests, signposts were installed such that the fields from adjacent signposts overlap. Each signpost periodically transmits a 16-Bit digital code using CFSK at 79.86 MHz. Each code includes a 7-Bit North code and a 7-Bit East code. Along any street, only the North or the East codes of adjacent signposts change, thereby creating a "chain." In the AVM equipped vehicle, a simple processor compares the codes received from adjacent signposts and determines if they "chain." By (1) identifying the code from the closest signpost, (2) comparing the codes from the two chaining signposts, and (3) comparing the levels received from each signpost, the vehicle is able to determine its location at one of five locations between each signpost pair. The combination of (1), (2), and (3) above results in an 18-Bit Location Region Code (LRC) which is stored in the vehicle until transmission over a communication link to the base station. At the base station each LRC is assigned to an X-Y coordinate in a data base file. Upon receipt of an LRC from a vehicle, the base station performs a table look-up and generates a physical X-Y location for the vehicle. When signposts are layed out as in the Philadelphia tests, an average of six unique LRC's are generated for each signpost. Thus, ignoring the perimeter, the number of valid X-Y location coordinates is approximately six times the number of signposts. Hoffman signposts may be lid out in different patterns. This flexibility allows virtually any specified location accuracy to be achieved. During fixed-route tests, signposts were located approximately one mile apart along the route. These signposts cause an odometer on the vehicle to be reset. An algorithm based on the last received signpost code and the elapsed odometer distance, provided accurate vehicle location. Testing involved recording, on cassette, the LRC output from a vehicle unit, the output of a reference fifth-wheel, and a series of manual reference inputs. Off-line processing of these cassettes produced location error statistics. Test results, including the effects of a simulated 5 percent communication error rate are presented.

Media Info

  • Features: Figures; References;
  • Pagination: p. 543-548

Subject/Index Terms

Filing Info

  • Accession Number: 00184857
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Feb 3 1981 12:00AM