Patent Application
20100179849
The invention relates generally to intelligent transportation brokerage systems and more particularly, to backhaul analysis in transportation brokerage systems.
Recently, there has been an increasing interest in collaborative logistics in freight transportation industry. Typically, shippers and carriers have managed operations independently. A new trend emerging is to collaborate with other shippers and carriers, identify potential opportunities on a system level, and share benefit of integrated operation costs among partners.
In commercial transportation operations, one of the major wasteful expenditures is movement of vehicles or tractor-trailers with little or no cargo. Analysis of inter-fleet data shows many lost opportunities for identifying backhauling loads i.e. cargo that may have been moved by an otherwise empty trailer on a return trip from a delivery point to a home base.
Brokerage systems that facilitate matching of load sharing and backhaul opportunities currently do not incorporate monitoring of real-time, geo-based information, and analysis of geo-based information from all brokerage participants. Currently, transportation brokerage systems match loads to participating partners either through individual driver's use of kiosks located at various stops of vehicles, or other brokerage services. Lack of automation results in vehicles pulling empty/partial cargo despite a potential for collaborations.
Therefore, there is a need for an improved, automated brokerage system for identifying backhaul opportunities and address one or more aforementioned issues.
In accordance with an embodiment of the invention, a method for identifying backhaul opportunities is provided. The method includes receiving data corresponding to a position, time and status from multiple vehicles. The method includes generating a database of trips made by the multiple vehicles based on the data received, wherein each of the trips is identified with a start point and an end point. The method also includes determining similarity between the trips to cluster the trips, wherein said determining comprises comparing a spatial distance between respective start points and respective end points with a pre-determined distance criteria. The method further includes identifying a frequency of each of the trips clustered made by the plurality of vehicles, wherein each of the trips are actual routes traveled between respective start points and respective end points. The method also includes generating a model for historical vehicle movement based on the identified frequent trips. The method also includes identifying multiple backhaul opportunities for a proposed empty trip using the model generated. The method further includes ranking the backhaul opportunities based on a frequency and cargo status of trips intersecting the proposed empty trip.
In accordance with another embodiment of the invention, a system for identifying backhaul opportunities is provided. The system includes a central data server to receive data corresponding to a position, time and status from multiple vehicles. The system also includes a processor configured to perform steps of generating a database of trips made by the multiple vehicles based on the data received, wherein each of the trips is identified with a start point and an end point. The processor is also configured to determine similarity between the trips to cluster the trips, wherein said determining comprises comparing a spatial distance between respective start points and respective end points with a pre-determined distance criteria. The processor is also configured to identify a frequency of each of the trips clustered made by the plurality of vehicles, wherein each of the trips are actual routes traveled between respective start points and respective end points. The processor is further configured to generate a model for historical vehicle movement based on the identified frequent trips. The processor is also configured to identify multiple backhaul opportunities for a proposed empty trip using the model generated. The processor is further configured to ranking the backhaul opportunities based on a frequency and cargo status of trips intersecting the proposed empty trip.
In accordance with another embodiment of the invention, a system for identifying backhaul opportunities is provided. The system includes multiple vehicle hubs that generate periodic status messages comprising location, time, cargo status and trigger event for a plurality of vehicles, each vehicle hub comprising a transmitter coupled electronically to the hub that broadcasts the status messages. The system also includes a central server configured to receive and store the status messages from the plurality of vehicle hubs in a telemetry database. The system further includes a trip extraction module configured to extract data from the telemetry database and generate a set of trips from the extracted data. The system also includes a characterization module configured to identify similar trips identified by the trip extraction module and to identify a frequency for each similar set of trips. The system further includes a backhaul matching module configured to match an empty trip of one of the vehicle hubs with one or more of the similar set of trips identified by the characterization module to identify a backhaul opportunity.
These and other features, aspects, and advantages of the present invention will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:
As discussed in detail below, embodiments of the invention include a system and method for identifying backhaul opportunities. The system and method provide an algorithm for near real time detection of backhaul opportunities in collaborative transportation systems, thereby reducing wasteful transportation behavior. There are two types of collaborative transportation systems. One is a centralized, on-demand transportation management system that includes commercial service providers. The other is a decentralized, online brokerage system that provides a marketplace for private transportation participants, to exchange information, respond to demand and supply fluctuations, and optimize operation costs.
In order to enable near real-time detection of backhaul opportunities, historical data obtained from a large-scale asset-tracking telematics system spanning an entire continent (North American) has been analyzed. Firstly, the technique identifies frequent trip routes from the telematics dataset in order to determine patterns in vehicle or freight movement. The technique further identifies backhaul opportunities based on frequency of trip routes that overlap, and status of cargo for respective routes.
It should be noted that embodiments of the invention are not limited to any particular processor for performing the processing tasks of the invention. The term “processor,” as that term is used herein, is intended to denote any machine capable of performing the calculations, or computations, necessary to perform the tasks of the invention. The term “processor” is intended to denote any machine that is capable of accepting a structured input and of processing the input in accordance with prescribed rules to produce an output. It should also be noted that the phrase “configured to” as used herein means that the processor is equipped with a combination of hardware and software for performing, the tasks of the invention, as will be understood by those skilled in the art.
P={lat,lon,t,e} (1),
wherein lat is latitude, lon is longitude, t is timestamp and e is an event code or status. Non-limiting examples of an event code are ‘trip start’, ‘trip end’, ‘door open’, ‘door close’, ‘cargo loaded’, ‘cargo empty’, ‘lost GPS signal’, and ‘low battery’. In operation, a remote hub 18 sends a ‘trip start’ event message when the vehicle 12 starts a trip and a ‘trip end’ event when the vehicle stops moving, as referenced by numeral 38. Data points correspond to a specific latitude and longitude. In order to generate or extract trips made by vehicles 12, the trips are represented by message sequences between consecutive ‘trip start’ events and ‘trip end’ events, as referenced by numeral 42. Hence, a trip T may be defined as:
T={P
i,(Pi+1, . . . ),Pj} (2)
wherein Pi and Pj are consecutive ‘trip start’ events and ‘trip end’ events respectively, and (Pi+1, . . . ) are possible intermediate non-trip-related event messages. Intermediate messages are critical in differentiating trips. Given the sparse nature of some telematics data, most of the trip data will have only start and end points, with no information about which route the asset takes on the trip. Locations of the intermediate messages provide additional information about a trip and can be useful to differentiate trips with different routes.
Due to the noisy nature of telematics data, heuristics rules can be added to the trip extraction process. In an ideal scenario, a ‘trip start’ message and a corresponding ‘trip end’ message appear in pairs, thus defining a trip. However, in practice, ‘trip start’ or ‘trip end’ messages may mismatch. For example, a ‘trip start’ message may be followed by another ‘trip start’ message, or a ‘trip end’ message may not have a corresponding ‘trip start’ message. In another embodiment, there may be multiple ‘trip start’ and ‘trip end’ messages missing. This can be inferred by checking if time duration between a ‘trip start’ and ‘trip end’ message exceeds a certain time threshold, say 3 days. In yet another embodiment, a ‘trip start’ message and a corresponding ‘trip end’ message are sent from the same location. This usually occurs when a vehicle has traversed a short distance roundtrip. The trip extraction algorithm based on the above heuristic rules will sequentially process the GPS data stream, looking for consecutive “trip start” and “trip end” message pairs. The duration and distance between the consecutive “trip start” and “trip end” message pairs are calculated to filter out exceptional cases.
In one embodiment, point 1 (42.3463, −71.0974), point 2 (42.3464, −71.0975), and point 3 (42.3460, −71.0976) all refer to the same location i.e. “Fenway park” in Boston, Mass. It should be noted that the coordinates are equivalent to a latitude and longitude of a particular location. Therefore, to determine if a start and end locations of two trips are spatially similar, a small radius d is used to address the fuzziness of locations represented by global positioning coordinates. The radius d may also be referred to as a predetermined distance. Nodes 62 represent significant locations such as, but not limited to, distribution centers, stores, vendors, and maintenance facilities. Links 64 are frequent trips going between these nodes and a thickness of each of the link 64 indicates frequency of the trips, wherein greater the thickness, higher is the frequency.
To calculate the actual route, a geometric network is constructed using software that allows for calculating routes and modeling the historical flow of monitored resources throughout a roadway network. In an exemplary embodiment, an ArcGIS software is employed. ArcGIS software is commercially available and well-known in the art. More details of the software may be obtained in internet website www.esri.com. Edges are weighted based on the cost or estimated travel-time to traverse each edge. The route with the minimal traversal time is often, but not always, considered the most likely traversed route. Additional information determined from analysis of telematics data can help validate the accuracy of the predicted route. The reasonableness of a proposed route is determined by comparing the estimated travel time to the actually observed duration. In another embodiment, to further improve accuracy, intermediate messages are analyzed. These are event-based messages beyond start/end of trip information, such as “door open”, or “cargo loaded.” Since each frequent trip is comprised of a set of individual trips, the in-transit messages from each trip assists in determining the frequently traversed routes. The frequent routes thus derived are used to create a model of historical freight movement.
In order to generate a model for historical vehicle movement, variables need to be associated with appropriate geographical locations and routes. Exemplary variables include cargo status and frequency information. It should be noted that other variables related to temporal information such as, extent of time collaboration or load sharing that may occur, also may be employed. Furthermore, location of distribution centers where trucks may be physically loaded and unloaded may be critical. As used herein, cargo status is defined as the ratio of full trips to total trips, and is recorded as a Boolean, specifically, 1=cargo_status, 0=no cargo_status. A value of 0 indicates that vehicles that traveled along that route were empty. Similarly, if cargo_status=1, then all the vehicles traveled full. A mean cargo_status of 0.5 would indicate half of the vehicles traveling that route as empty and half were full. Knowledge of cargo status is crucial in assessing backhaul opportunities i.e. matching between empty trips and full trips occurring in the same direction. A similar process occurs for route frequency, except that the frequency for each route is initialized based on number of trips clustered together. The frequency of these trips weights a backhaul opportunity in determining likelihood that a collaborative match may occur within temporal restraints. In the model, each road segment is embedded with cargo status and frequency information for each direction of travel. To determine cargo status and frequency at specific route segments, routes that overlap need to be combined.
To determine cargo status and frequency features, a roadway network is segmented based on overlap and direction of travel. The frequency and cargo status is calculated for all road segments. A route segmentation algorithm creates a new route Rnew (110) which is defined by intersection of the route 102, denoted by R1 and route 104, denoted by R2:
Rnew=R1∩R2 (3)
Three routes are created from the two original routes 102 and 104 to distinguish between the three different segments Rnew, R1, and R2. The three segments 102, 104, and 110 respectively are: (1) where R1 and R2 intersect, frequency is two and cargo status is one-half; (2) disjoint set of R1 and Rnew, this defines section of R1 where R1 and Rnew do not overlap. In this case, frequency is one and cargo status is one and (3) the disjoint set of R2 and Rnew, with frequency is one and cargo status is zero.
The different states that exist between the intersecting and non-intersecting segments require dividing two routes into three when a non-empty intersection occurs. Such segmentation is performed for every route in the domain. For each intersection 112, variables are determined at the corresponding geographic location. Examples of these variables include frequency and cargo status at each location along the roadway network. Initially, each frequent route has a variable frequency that is initialized by the number of trips that were clustered into that frequent route. An additional variable cargo_status is initialized to the ratio of full to empty trips for each direction along the road network.
In order to detect backhaul opportunities, the historical network 120 is queried based on a proposed empty trip. Two inputs are required for determining backhaul collaboration opportunities: accurate determination of historical vehicle or freight movement patterns and current intentions of a distributor. In an example, consider a situation in which a driver has just delivered a shipment to a Virginia/North Carolina border. A driver now needs to return towards a distribution center outside the Richmond metropolitan area. The cost of fuel, driver's pay and vehicle overhead make it expensive to drive a long distance without pulling any revenue generating cargo. A backhaul opportunity exists if the empty trip intersects with a frequently traveled route. Such potential intersections are identified based on the historical vehicle movement patterns, as described above, (
At the intersection, likelihood of collaboration is determined based on frequency of the route and percentage of full cargo loads on the route, also defined as cargo status. The higher the frequency, the more likely backhaul collaborations may occur within time restrictions, since waiting for potential loads is often expensive. Additionally, a high percentage of full loads or cargo status is an indicator of a greater demand for hauling goods into a particular region. Furthermore, basic economic theory states that higher demand results in higher prices. Therefore, high demand for freight transportation in the same desired direction should result in substantial payoff and therefore increase motivation for collaboration.
If Historical_network∩Empty trip=Φ, then
Print “No viable opportunity”
Else
RANK=frequency*cargostatus
End (4)
Thus, the backhaul opportunities are ranked as per above criterion. If the empty trip does not intersect with a frequently traveled route, then historical trends indicate that it is more difficult to collaborate on backhaul. In such cases, visual inspection of historical vehicle movement patterns may assist a fleet manager in identifying cost-saving behavioral changes. The changes may include deviations from shortest route to allow for collaboration to occur. A product of frequency and cargo status determines a ranking of viability of the backhaul opportunity. In a particular embodiment, external variables that impact backhaul collaborations also referred to as risk factors may be considered. Some non-limiting examples of the risk factors include travel time and network constraints. In an example, a high variance in travel time correlates to higher risk. In another example, avoiding traveling in a certain area due to environmental or safety constraints correlates to higher risk.
The various embodiments of system and method to identify backhaul opportunities described above thus provide near real-time automated detection by using a large telematics network tracking hundreds of thousands of assets. The system and method facilitate automated business partner discovery and multi-hop schedule recommendations. The technique also benefits larger fleets and improves freight transit efficiency, thus reducing number of vehicles traveling with empty cargo. This further reduces amount of CO2 and NOx emissions produced by the vehicles. By identifying backhaul collaborative opportunities, a number of empty miles can be reduced, saving money on fuel, salary and vehicle costs and reducing emissions.
It is to be understood that not necessarily all such objects or advantages described above may be achieved in accordance with any particular embodiment. Thus, for example, those skilled in the art will recognize that the systems and techniques described herein may be embodied or carried out in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other objects or advantages as may be taught or suggested herein.
Furthermore, the skilled artisan will recognize the interchangeability of various features from different embodiments. For example, the use of a GPS to receive position data with respect to one embodiment can be adapted for use with a processor configured to assess one or more risk factors described with respect to another. Similarly, the various features described, as well as other known equivalents for each feature, can be mixed and matched by one of ordinary skill in this art to construct additional systems and techniques in accordance with principles of this disclosure.
While the invention has been described in detail in connection with only a limited number of embodiments, it should be readily understood that the invention is not limited to such disclosed embodiments. Rather, the invention can be modified to incorporate any number of variations, alterations, substitutions or equivalent arrangements not heretofore described, but which are commensurate with the spirit and scope of the invention. Additionally, while various embodiments of the invention have been described, it is to be understood that aspects of the invention may include only some of the described embodiments. Accordingly, the invention is not to be seen as limited by the foregoing description, but is only limited by the scope of the appended claims.
The present patent application claims priority from provisional patent application Ser. No. 61/143,508, filed Jan. 9, 2009, the disclosure of which is hereby incorporated by reference in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 61143508 | Jan 2009 | US |
